| 20080220402 | METHOD FOR CREATING A FAITH BASED LEGACY MEMORIAL | September, 2008 | Chaveriat |
| 20060088812 | Apparatus and method for automatically analysing a filled in questionnaire | April, 2006 | Slijp et al. |
| 20090269730 | SIMULATION AUTHORING TOOL | October, 2009 | Nguyen et al. |
| 20080261191 | Scaffolding support for learning application programs in a computerized learning environment | October, 2008 | Woolf et al. |
| 20080319560 | AUTOMATED RECIPE COMPOSER | December, 2008 | Fayez |
| 20010049087 | System and method of distance education | December, 2001 | Hale |
| 20070243508 | Diabetic teaching process | October, 2007 | Fitzjarrald Jr. |
| 20070154870 | Customizable wireless education or occupational therapy tool having a switch unit and a transmitter unit | July, 2007 | Ongkojoyo |
| 20060257828 | Mathematical game | November, 2006 | Tsui Collins |
| 20070048692 | Conceptual sorting ability assessment and associated methods | March, 2007 | Kemp et al. |
| 20030170605 | Vapor deposited writing surfaces | September, 2003 | Long et al. |
Under 35 U.S.C. §119(e)(1), this application claims the benefit of prior co-pending U.S. Provisional Patent Application Nos. 61/081,262, Cognitive Amplification For Contextual Game-Theoretic Analysis Of Courses Of Action Addressing Physical Engagements, by Schlabach et al., filed Jul. 16, 2008, and 61/120,217, Cognitive Amplification For Contextual Game-Theoretic Analysis Of Courses Of Action Addressing Physical Engagements, by Schlabach et al., filed Dec. 5, 2008, both incorporated herein by reference.
Under paragraph 1(a) of Executive Order 10096, the conditions under which this invention was made entitle the Government of the United States, as represented by the Secretary of the Army, to an undivided interest therein on any patent granted thereon by the United States.
Battles are often decided based upon which side first reaches key terrain. It is unlikely to conceive a Union victory at Gettysburg if General Buford's Cavalry had required a few extra hours to arrive at the high ground south of town. And yet, using widely-accepted guidelines, it is estimated that over half of a modern Brigade's battle-preparation time is consumed by Brigade and Battalion staff planning, before the Company Commanders even receive the mission and start their own planning. So, any reduction in the amount of time a staff requires for planning results in the unit being faster into action.
The Battlefield Terrain Reasoning Awareness, Battle Command (BTRA-BC) Battle Engine (BBE) offers the potential for significant reductions in the amount of time a staff requires for battle planning. (See also Appendix F for a Glossary of Terms). The prototype BBE application represents an implementation of “Cognitive Amplification” for Contextual Game-Theoretic Analysis. (See also Appendix B for a detailed description of the BBE). As such, it presents a fully functional capability to explain and explore the advantages offered by this approach to military decision making. The BBE cognitively amplifies the ability of battle staff planners to conduct Intelligence Preparation of the Battlefield (IPB) and the Military Decision Making Process (MDMP). Consequently, a “human-computer reasoning team” develops and analyzes tactical Courses of Action (COA's) much faster than humans alone and better than computers alone.
Refer to FIG. 2 illustrating the relationship 200 defining Cognitive Amplification as may be employed in select embodiments of the present invention. Data 201 such as may be employed in the BBE is input to provide Information 202 usable by a human. From this information, Knowledge 203 is gained by both the system and a human user. This is the “science” of the process. The Knowledge 203 gleaned from manipulating the Information 202 via a computer increases the Knowledge 203 of the user that enables the user to employ the art (combining the human's experience, training and background with the Information 202 from the “science”) to yield Wisdom 204 that enables informed efficient planning. There is a distinction between Cognitive Amplification (CA) and Artificial Intelligence (AI). Unlike many AI efforts, CA both retains a human “in the loop” and assigns the human responsibility for the loop. In military applications, for example, the time saved in conducting IPB and MDMP results from computers executing “Military Science” significantly faster than a human while a human expert properly employs “Military Art” in a manner unable to be implemented with a computer. In all cases a human directs a computer's “battlefield reasoning,” thus the resultant “automated cognition” is that of a human, albeit assisted in the science of it by a computer.
Conventionally, there are three fundamental approaches to executing the Military Decision Making Process (MDMP) either in the field use or in research environments:
Human cognition without Computer Amplification. This is the dominant field approach, in which human experts conduct a significant majority of the MDMP cognitive processes manually. Computer Command and Control (C2) systems are used primarily to record results of human cognitive analysis or to aid in visualizing a human-conceived state of battle during a war gaming session. There are several prototypes that may assist in developing detailed plans, but these still require a human expert to input a COA concept in a Mission, Enemy, Terrain, Troops, and Time (METT-T) game context. Much time and intellectual energy is consumed in identifying, developing, and analyzing the variables that comprise Friendly COA's (FCOA's) and Enemy COA's (ECOA's). Advantages of select embodiments of the present invention over Human Cognition without Computer Amplification are speed, precision, and comprehensiveness. A human expert can develop appropriate COA concepts, in context, in a few minutes, the automated Terrain Informed War Game Model employed in select embodiments of the present invention conducts a war gaming analysis in under a second whereas an experienced human staff typically takes an hour or longer. (See also Appendix D for detailed explanation of the Terrain Informed War Game Model). Further, the precision of the war gaming results from employing select embodiments of the present invention is significantly greater than the roughly approximated attrition estimates from a typical manual war gaming session. Finally, select embodiments of the present invention achieve an extraordinary war gaming speed that facilitates a significant increase in comprehensiveness. Human war gaming sessions typically analyze only one battle in detail, i.e., an initially selected FCOA against a most likely ECOA. If the planners conduct any war gaming beyond that (e.g., against other ECOA's), it is typically highly abbreviated. Select embodiments of the present invention can fully war game analyze a large set of FCOA's against a large set of ECOA's in mere seconds.
FOX FOX is a family of research prototypes originally developed by one of the present inventors at the University of Illinois at Urbana-Champaign. FOX uses a “fast abstract” War Game Model to estimate the results of battle. The battle estimates coming out of FOX are inconsistent in their realism and reliability due to immature modeling of weapon and terrain effects for combat attrition. FOX also has a relatively crude mechanism for desired end state. This led to FCOA evaluations that had a large variance compared to a user's actual desired end-state. Select embodiments of the present invention offer a significantly improved, Terrain Informed, force-articulated combat attrition model, to include a set of end state options for a user's consideration. Thus select embodiments of the present invention closely model a user's actual needs. Since the FCOA's produced by the genetic algorithm (GA) employed in select embodiments of the present invention converge upon the end state evaluation criteria, the improved end state mechanism provides this invention significantly with more directability and agility than FOX. (See also Appendix A for a detailed discussion on genetic algorithms). Select embodiments of the present invention also have significantly improved COA Variables compared to FOX, also improving relative directability. Finally, FOX focuses on just the war gaming aspect of the MDMP process. Select embodiments of the present invention provide significantly expanded support to a greater subset of the MDMP process. Select embodiments of the present invention add a front end mission analysis, increased FCOA evaluation support tools, support to a user's initial decision via a risk deprecation analysis, and improved and expanded post-decision planning via support for the IPB Event Template. (See also Appendix E for a detailed discussion of deprecation analysis). This results in a comprehensive game-theoretic support system in which it is easier for a user to fully explore implications of a current METT-T scenario.
Modeling and Simulation (M&S) systems. The military M&S community maintains high-fidelity combat models suitable for use with FOX and select embodiments of the present invention.
These models support controlled MDMP experiments, and the training community often uses these models to support collective battle staff training, including training staff on MDMP procedures. The advantages of select embodiments of the present invention over an M&S approach are speed, usability, and game-theoretic comprehensiveness. The M&S models typically take at least an hour to run a single battle simulation, often requiring more time. Select embodiments of the present invention simulate battle and produce a battle script in a few milliseconds. The M&S systems also usually require a fair number of experts to execute a battle simulation. Even though select embodiments of the present invention are designed for collaborative use by several users (planners), it can be directed easily by one person, making it more usable than the most usable of the M&S systems. Finally, M&S systems are so resource intensive, particularly with respect to time, that they are seldom used to develop a game-theoretic context map of the best FCOA's and best ECOA's available in an engagement scenario. Select embodiments of the present invention excel at providing comprehensive game-theoretic (ECOA v. FCOA) context.
SHAKA (later renamed FOX), a prototype of the present invention, was impractical due to its inability to reason and plan within context, particularly regarding effects of terrain, in particular effects impacting combat tactics. The advantages of select embodiments of the present invention over FOX are realism, directability, and full game-theoretic feature support. FOX uses a relatively immature model for terrain effects. Select embodiments of the present invention leverage sophisticated terrain analysis products produced by the U.S. Army Topographic Engineering Center (TEC). In select embodiments of the present invention, these products include an “articulated” Modified Combined Obstacle Overlay (MCOO), with a Braswell reference of combat terrain effects for each maneuver corridor represented in the MCOO. (See also Appendix B for a detailed discussion of the Braswell Index). FOX also uses an immature model for combat attrition, whereas select embodiments of the present invention leverage the well-established Dupuy QJMA methodology. Select embodiments of the present invention offer significant improvement in the terrain and attrition models leading to highly realistic estimates.
Select embodiments of the present invention deploy a responsive, abstract, Terrain Informed war gaming engine that reasons within a comprehensive context particularly suited to making decisions in a time-constrained scenario, such as during battle preparations. Select embodiments of the present invention are more realistic than FOX and more “user friendly” than M&S systems. By exhaustively comparing the multitude of variables that comprise FCOA's and ECOA's, select embodiments of the present invention permit a user to expend intellectual energy considering the effect of these variables rather than trying to identify them in the first place.
FIG. 1 is a flow chart that summarizes a process employed by select embodiments of the present invention.
FIG. 2 illustrates the relationship defining Cognitive Amplification as may be employed in select embodiments of the present invention.
FIG. 3 is a map overlaid with paths resulting from employing the Braswell Index with select embodiments of the present invention.
FIG. 4 is a screen print illustrating visualization of a battle snapshot as may be made available when employing select embodiments of the present invention.
FIG. 5A is an example output of the BTRA-BC MCOO-Maker accomplished by repeatedly using a routing algorithm to conduct a network-pulse analysis of mobility corridors, as employed in select embodiments of the present invention.
FIG. 5B represents an output of the network-pulse analysis of mobility corridors that yields a set of V-lanes that are logically-parallel routes through the network of mobility corridors, as employed in select embodiments of the present invention.
FIG. 6A illustrates an example of the battlefield-physics process the METT-T Parser uses to identify all possible instances of an important COA variable, Unit Boundaries, as employed in select embodiments of the present invention.
FIG. 6B depicts Pascal's Triangle for binomial expansion, a concept used in developing select embodiments of the present invention.
FIG. 7 is a screen print of a Graphic User Interface (GUI) for a Defensive COA selection as may be displayed by select embodiments of the present invention.
FIG. 8 is a screen print of a GUI for an Offensive COA selection as may be displayed by select embodiments of the present invention.
FIG. 9 lists component sub-processes within a Terrain Informed War Game Model as may be employed with select embodiments of the present invention.
FIG. 10 depicts a screen that may be used to establish criteria for a Desired End State as may be employed with select embodiments of the present invention.
FIG. 11 shows a subset of the diagram in FIG. 1 that emphasizes operation of the FCOA Evaluator in directing the Terrain Informed War Game Model to engage an FCOA chosen from the FCOA Candidates against all of the ECOA Candidates as may be employed with select embodiments of the present invention.
FIG. 12 is a screen print representing the predicted performance of a select FCOA given selected criteria for each of three representative ECOA's as may be employed with select embodiments of the present invention.
FIG. 13 depicts the use of an iteration cycle that both manual and automated optimization techniques may employ in select embodiments of the present invention.
FIG. 14 is a screen print of a page of a table summarizing a deprecated risk analysis as may be employed with select embodiments of the present invention.
FIG. 15 is a screen print of an example list of evaluation results showing the use of Red-Amber-Green color highlighting as used in select embodiments of the present invention.
FIG. 16 is a screen print example for a set of FCOA's generated by the genetic algorithm (GA) that also provides a button for performing a Pareto Analysis, as used in select embodiments of the present invention.
Select embodiments of the present invention envision a process that employs computer amplification of a human expert's cognitive reasoning to generate and analyze military Courses of Action (COA's) in a terrain informed, game-theoretic context. This cognitive amplification process leverages the institutionalized military procedures of Intelligence Preparation of the Battlefield (IPB) and the Military Decision Making Process (MDMP) to ensure the human expert and the computer share a common reasoning framework. Select embodiments of the present invention divide the cognitive tasks so that the human expert executes and directs military “art.” Military art requires judgment and decisions that are difficult for a computer to address in context thus the need for human judgment. In contrast, a computer is best suited to execute cognitive tasks associated with military “science.” Military science typically requires amounts of correlation and calculation that would be intolerable for a human expert to perform, especially under typical time constraints involved in battlefield scenarios.
Select embodiments of the present invention are designed to simultaneously provide: fast-abstract war gaming; realistic estimates of the combat effects of terrain; realistic combat attrition estimates; comprehensive integration of the MDMP and IPB doctrinal processes; easy usability; and computer reasoning in harmony with, and at the direction of, human users. Before development of select embodiments of the present invention some of these design goals were at odds with one another.
Select embodiments of the present invention, functionally described as human-computer cognitive amplification, provide a decision support system. In military applications, this system may be employed by a tactical commander, typically by his battle staff. The commander may use embodiments of the present invention to readily develop a comprehensive understanding of the relative strengths and weaknesses of candidate Friendly COA's (FCOA's) within the game-theoretic context of likely and dangerous Enemy COA's (ECOA's). That is, select embodiments of the present invention applied to military applications improve a commander's ability to decide, in a timely manner, optimum tactics to employ in battle.
Refer to FIG. 1, a flow chart summarizing a process employed by select embodiments of the present invention. In select embodiments of the present invention, semi-automated procedures that are steps in the process closely adhere to cognitive processes used for decision making (such as the military's doctrinal MDMP) and for employing intelligence (such as the military's doctrinal IPB).
The first four procedures (steps) of the process of FIG. 1, the articulated Modified Combined Obstacle Overlay (MCOO) and Braswell Index 101, the Enemy Order of Battle (OB) 102, the Friendly Order of Battle 103 and the Missions and Postures 104, are data inputs. The remaining steps comprise a mixture of automated procedures and human expert inputs. The process represented in FIG. 1 greatly increases the efficiency of decision making and use of background (“intelligence”), both heretofore depending almost exclusively on human cognition.
In select embodiments of the present invention inputting the Articulated MCOO and Braswell Index 101 establishes a game board upon which an automated planner, such as a Terrain Informed War Game Model 112, may develop attrition estimates of a battle between an FCOA and an ECOA. In select embodiments of the present invention, this input includes a Modified Combined Obstacles Overlay (MCOO), a basic product from the U.S. Army's Intelligence Preparation of the Battlefield (IPB) doctrinal process. An MCOO illustrates the maneuver options for units (tokens on the game board) in a given engagement (battle) situation by identifying obstacles to aggregated token (unit) movement, the mobility corridors 301 (FIG. 3) between those obstacles, and the logical groupings of mobility corridors 301 to form what were historically termed Avenues of Approach (AA's). Operational planners typically have used the MCOO to structure maneuver options for both offensive and defensive COA's.
Select embodiments of the present invention employ a modified (articulated) MCOO developed from a software application called the MCOO-Maker. The MCOO-Maker is part of the BTRA-BC program developed at the Topographic Engineering Center (TEC) of the U.S. Army Corps of Engineers Engineer Research and Development Center (ERDC). The BTRA-BC MCOO-Maker uses a logical partition of an area of operation (AO) known as the Braswell Index. In select embodiments of the present invention, the MCOO-Maker builds upon the Braswell Index to establish a logical infrastructure to support the development of Avenues of Approach (AA's) and Lines of Defensible Terrain (LDT's), consistent with the IPB doctrinal product, the Modified Combined Obstacle Overlay (MCOO). The products from the MCOO-Maker are in greater detail than a typical human-developed MCOO, thus this product is termed an Articulated MCOO. The articulated detail helps a computer explicitly reason through issues that human experts implicitly understand. As an example, experienced planners implicitly know how many subordinates could attack abreast in a given AA, whereas a computer must explicitly tag each AA with its ability to support side-by-side formations of subordinate units. Otherwise, the BBE could not match the ability of a skilled planner to develop COA's with feasible formations in each AA.
FIG. 3 shows one view of the Braswell Tactical Spatial Index (Braswell Index) that enables the BBE to “abstract away” detailed terrain features while retaining the terrain effects upon combat attrition modeling. The darker polygons 302 represent major obstacles to unit movement, while the network of heavy lines 301 represent the center line of the mobility corridors 301, i.e., the lines 301 are a one dimensional (1D) representation of a two dimensional (2D) corridor between every pair of obstacles 302. Combat actions are usually compartmented within those mobility corridors 301, so the BBE's attrition calculations may reference just the smaller data of combat effects for the pertinent mobility corridor 301. For example, a mobility corridor 301 with excellent Cross-Country Mobility (CCM) will increase the (otherwise) combat power of an aggressor. The BBE needs only to import that corridor's CCM combat multiplier rather than that corridor's detailed, geo-rectified CCM information. The BBE retains the feature class key of each mobility corridor 301 to support later visualization in a GIS program such as ArcMap.
Refer to FIG. 3, an annotated map 300 indicating with center lines 301 that indicate the center of the polygon representing the entire mobility corridor 301. In select embodiments of the present invention these maps 300 result from employing the Braswell Index. In select embodiments of the present invention, the Braswell Index establishes a connected network of mobility corridors 301 in and around obstacles 302 to token (unit) movement, consistent with military IPB MCOO doctrine. This abstraction is highly useful because combat between opposing forces typically occurs in an area between two obstacles 302. The Braswell Index identifies and establishes boundaries for this area in order to develop mobility corridors 301. These non-overlapping mobility corridors 301 (i.e., no center lines 301 cross) become a useful spatial index for abstracting (reducing) highly detailed terrain analysis information into compartments (corridors) likely to host engagements such as firefights (small battles).
The map 300 employs a visualization of “polylines” 301, 301A, certain center lines 301A of which bisect what one may consider “open” mobility corridors 301 (only the center lines 301 of which are visible) and others 301 of which represent “pinched” mobility corridors (hereafter all mobility corridors are generically identified as mobility corridors 301). In essence, this connected network of mobility corridors 301 defines the maneuver possibilities for units that want to move through an area. The mobility corridors 301 also define the physical boundaries that would likely contain local engagements, such as firefights. The polylines 301, 301A are the geo-rectified, abstract representations of the associated mobility corridors 301, 301A. This abstraction enables loading a game board into a computer's RAM, rather than the hard drive, literally resulting in a thousand fold increase in speed. This enables the Terrain Informed War Gamer 112 to fight fast, but still retain realism in combat attrition estimates.
In select embodiments of the present invention, the Braswell Index adds further value by providing an organizational and referencing scheme (an index) for a large catalogue of terrain analysis map overlays, such as cross-country mobility (CCMA, soil composition, vegetation, elevation, and the like. The Braswell Index allows for an efficient characterization of the combat effects of terrain in each mobility corridor 301. This improves efficiency because the fully geo-rectified dataset of terrain characteristics is very large, and thus inefficient for calculations that leverage all of that data. The abstracted data using the Braswell Index retains the meta-information required for effective modeling, while also supporting significantly more efficient calculations.
For example, the mobility corridor 301 with a narrow choke point represented by 301B (as compared to the open mobility corridor 301C) and highly-restricted cross country mobility (CCM) due to a large area obstacle 302A would greatly favor a defender over an attacker. In essence, this local terrain situation would increase the (otherwise) combat power of the defender, and decrease the (otherwise) combat power of the attacker. Another mobility corridor 301 might have the exact opposite combat effects, depending upon the local terrain situation. This Braswell Index of terrain analysis products enables the MCOO-Maker to send just abstracted combat effects rather than a much larger catalogue of terrain analysis products. In select embodiments of the present invention, the abstracted combat effects for each indexed mobility corridor 301 are a set of quantified multipliers that may be applied later to the relative combat powers of opposing forces engaged in that mobility corridor 301.
In select embodiments of the present invention, this highly abstracted index of terrain effects enables the Mission, Enemy, Terrain, Troops, and Time (METT-T) Parser 105 to load a still-realistic representation of the battlefield (game board) into a computer's basic memory (RAM), rather than onto a hard-drive. (See also Appendix C for a detailed description of the METT-T Parser). Employing “RAM only” results in simulations that run approximately a thousand fold faster than simulations that must access a hard drive. Computer access to RAM is measured in microseconds, whereas access to hard drives is measured in milliseconds due to the mechanical constraints of the hard drive.
METT-T is an institutional acronym used to denote the battlefield situation by listing the fundamental military components of that particular situation. In select embodiments of the present invention, METT-T Battle Context Mapping improves both speed and quantity employing a human-computer set of procedures (FIG. 1) that enables strong exploration of the game-theoretic dynamic of a pending engagement consistent with military MDMP and IPB doctrine. The mapping (survey) of this game-theoretic dynamic enables a computer to have an articulated appreciation of battle context in a knowledge format easily understandable by human experts directing the analysis. Using terminology from the study of chaos, select embodiments of the present invention allow a computer to acquire “emergent intelligence” about the engagement scenario (battlefield).
Refer to FIG. 4, a screen print 400 of a visualization of an engagement (battle) snapshot using the abstracted Braswell Index as the game board 403. The network edges (line segments) 401 behind the light colored boxes represent the set of Mobility Corridors 301 that constitute Lines of Defensible Terrain 2 (LDT-2). The dark blocks 402 represent attacking forces that have bypassed the defense on LDT-2, and are using other component Mobility Corridors 301 on their V-Lanes 501 to proceed to their attack objectives towards the right.
Refer to FIG. 5A, in which the BTRA-BC MCOO-Maker develops a highly articulated, but still doctrinal MCOO by repeatedly using a routing algorithm to conduct a network-pulse analysis of the Mobility Corridors 301. In select embodiments of the present invention, this analysis uses network routing since combat units (teams) may traverse the area of operation abreast, thus the displacement footprint of each unit must be considered. Furthermore, the routing analysis is “pulse” rather than continuous, since offensive combat units typically traverse an area once, in a predetermined formation.
Two examples of the additional battlefield physics information an Articulated-MCOO provides are Virtual (V) Lanes and Lines of Defensible Terrain (LDT's). A defender's LDT's are logically perpendicular to the attacker's V-Lanes 501 (FIG. 5), and constitute logical groupings of neighboring mobility corridors 301 upon which a coherent defense can be based. Again, a human expert can usually identify these at a glance, and implicitly reason through the battlefield physics of potential COA's. The MCOO-Maker provides an explicit representation of these LDT's, thus the computer can “reason” through the same battlefield physics as the expert uses.
Refer to FIG. 5B. V-Lanes 501, output from the Network-Pulse Routing Analysis, are explicitly more flexible than the Avenues of Approach (AA's) historically displayed on a classic MCOO. A trained human analyst who draws an AA on a MCOO does not annotate the AA by its ability to support two or more subordinate units attacking abreast. The experts that use the MCOO simply understand these implicit capabilities at a glance. The experts can reason through tactical Courses of Action (COA's) that consider the full capacity of that AA. A computer can not implicitly digest these AA capacities, so the BTRA-BC MCOO-Maker explicitly identifies the V-Lanes 501 that enable the computer to reason through the same battlefield physics that a human expert uses to identify potential COA's.
Refer to FIG. 5B. In select embodiments of the present invention, a network-pulse analysis of Braswell-established mobility corridors 301 yields a set of V-lanes 501 that are “logically parallel” routes through the network of mobility corridors 301 from the attacker's start line to the attacker's objective line. This is termed an articulated MCOO because it offers almost all the information of a doctrinal MCOO while providing information about the “battlefield physics” to “inform” the METT-T parser 105.
Refer again to FIG. 1. The METT-T Parser 105 and associated COA Variable Set 106, 107 provide a Terrain Informed articulation of the major elements of a user's (commander's) “abstracted” concept decision. The METT-T Parser 105 develops all possible battlefield physics instances for each COA Variable 106, 107, and arranges sets of instances to reasonably maximize neighborliness. This facilitates later FCOA optimization through the genetic algorithm (GA). Neighborliness is an informal term that describes the correlation between any two adjacent instances of COA Variables 106, 107 and their contributions to the final evaluation score of a solution when all other variables are controlled.
In select embodiments of the present invention, a Terrain Informed War Game Model 112 employs an “attrition model” to determine likely results of combat. Before implementation of the Terrain Informed War Game Model 112, the two effects of “fast-abstract” and “Terrain Informed” were almost mutually exclusive for use in combat simulations. Select embodiments of the present invention integrate the two. In turn, an attrition calculation based on the attrition model requires estimates of relative combat power of opposing forces. Estimates must appropriately account for the number and types of weapon systems on each side. Select embodiments of the present invention leverage the well-established Quantitative Judgment Method of Analysis (QJMA), described by Colonel (Ret.) Dupuy. Dupuy, Trevor N., Numbers, Predictions, and War: Using History to Evaluate Combat Factors and Predict the Outcome of Battles, Bobbs Merrill, Indianapolis, Ind., ISBN 0-672-52131-8, 1979. The QJMA provides a strong historical basis for assessing the relative powers of individual weapons.
Select embodiments of the present invention receive as input the relative weapons estimates from a BTRA-BC application termed the BTRA-BC Battle Engine Weapons Assessment and Calculation Tool (B-WAC7). The B-WACT implements the QJMA concept to develop a basic relative combat power for individual weapons and weapon systems that aggregate weapons. For example, a heavy battle tank usually has a main gun and several machine guns that work in a synergistic fashion. A user provides a weapon's characteristics to the B-WACT which then provides a QJMA relative combat power for that weapon. The B-WACT also enables a user to aggregate weapons into larger weapon systems that also receive a QJMA relative combat power. Finally, the B-WACT publishes lists of weapon systems as a data file as a possible input. Select embodiments of the present invention enable a user to quickly build OB files for enemy units, thus the aggregated relative combat power of the enemy units is strongly grounded in a QJMA weapons estimate output from the B-WACT.
The Unit Posture ratings influence the attrition calculations in the Terrain Informed War Game Model 112. For example, a “deliberate defense” emplaces weapon systems in protected firing positions prepared by combat engineers. Since those systems are now more combat effective, the defender receives a “multiplier.” A “hasty defense” typically does not have time to prepare such positions, and receives an appropriate multiplier based on the defender's ability to quickly take advantage of cover in terrain readily available in the immediate area.
The Unit Morale ratings supplement the Terrain Informed War Game Model 112. For example, U.S. forces in Operation Desert Storm enjoyed “Excellent Morale” by almost any historical standard. In contrast, the morale of the Iraqi Army ranged from “Good Morale” (for a handful of Republican Guards Units) to “Panicked Morale” for some poorly led, poorly trained, and poorly motivated reserve units. These panicked units were very hesitant to fully engage U.S. forces, so they did not employ their weapons to full potential. The Unit Morale setting allows the Terrain Informed War Game Model 112 to appropriately degrade the QJMA relative combat power.
The Superiority Toggle ratings supplement the Terrain Informed War Game Model 112. For example, if an enemy unit has local intelligence superiority due to a sympathetic populace, then its forces are assigned an appropriate combat multiplier in the Terrain Informed War Game Model 112.
The Game Time Slice ratings supplement the Terrain Informed War Game Model 112. For example, if a user sets the Game Time Slice to 18 minutes, then the Terrain Informed War Game Model 112 simulates unit movement through discrete displacements reflecting how far each unit can move in 18 minutes, assuming the unit is in maneuver mode. As explained below, after all units displace their appropriate 18-minute distances, the Terrain Informed War Game Model 112 updates attrition and other status information for all units.
The Missions and Posture 104 inputs from a user, together with the other three initial inputs 101-103 described above, provide game context to all subsequent battle reasoning. From the perspective of military science, those inputs 101 - 104 provide METT-T context.
In select embodiments of the present invention, the Friendly Commander decides upon a tactical Friendly COA (FCOA) 107 by selecting a variable instance for each of the variables in the set. The commander's staff typically analyzes a set of candidate COA's from which the commander decides which tactic(s) to employ.
A Friendly Commander rarely has direct knowledge of the Enemy Commander's tactical decision, thus during staff analysis the intelligence officer(s) estimate a logical, representative set of Enemy COA (ECOA) 106 options, against which the friendly staff ideally conducts a risk analysis for each FCOA in the FCOA candidate set 111.
The doctrinal military term for analyzing candidate FCOA's against a set of ECOA options is “war gaming.” In select embodiments of the present invention, the METT-T Parser 105 identifies all possible battlefield physics options for each of the pertinent COA variables 106, 107. Note that there are “dominate” variables (upon which others may depend) and “dominated” variables (which may depend on other variables). As a result, the Terrain Informed War Game Model Model 112 analyzes the effect of reasonable “tactical dynamics” a commander and staff may analyze.
Refer to FIG. 6 A, illustrating an example of the battlefield physics process the METT-T Parser 105 uses to identify all possible instances of a COA variable, Unit Boundaries. The Unit Boundaries variable is used in both Offensive and Defensive COA's. For example, if a user plans an attack in two columns using the Num Abreast COA variable, a user must decide where to place the internal unit boundary between the two columns. In FIG. 6A, the user has six possible locations (between V-lanes #1-#6 501) for internal unit boundaries that have been identified in the Articulated MCOO and Braswell Index input 101. Assuming a user selected the #2 instance of the Num-Abreast COA variable, a user must further select one of these six possible internal boundary options before his total COA selection is complete. If a user has five available subordinate units, there will be an associated set of possible Boundary Variable instances for each of the five possible instances of the dominating Num-Abreast COA variable. As illustrated in FIG. 6B, these sets rigorously follow Pascal's Triangle for binomial expansion, a fact useful in organizing a computer's memory (RAM) for this COA Variable.
The number of available subordinate units in the Friendly OB 103 and the number of available V-Lanes 501 input in the Articulated MCOO and Braswell Index input 101 affect the number of possible instances in the Unit Boundaries COA Variable. A user can not provide for an attack three-abreast if there are only two available V-Lanes 501 or two available units. The METT-T Parser 105 accounts for this logic.
In select embodiments of the present invention using similar battlefield physics logic, the METT-T Parser 105 establishes sets of all possible instances for each of the COA Variables, for both attacker and defender as explained fully below. For example an Offensive (Attacker's) COA Variable Set is assigned to the Friendly Forces to structure the FCOA variable set 107, while the Defensive COA Variable Set is assigned to Enemy Forces to structure the ECOA variable set 106. The METT-T Parser 105 also supports the opposite “binding” (Enemy forces in attack and Friendly forces in defense).
In the METT-T circumstance of this example screen, the enemy is in a defensive posture, has three available subordinate units, five V-Lanes 501, five LDT's, and five Task-Organizable Subordinates. From this input the METT-T Parser 105 populates the following ECOA Variables:
Total Unit Variables 701A: The METT-T Parser 105 establishes a set of ECOA Variables for the total unit. In the screen 700, options in the left panel 701 A change to reflect the ECOA selections:
Again referring to FIG. 7:
The variables for Offensive (attack) COA's are similar to those used in the Defensive COA's described above in FIG. 7. FIG. 8 represents variables for use in planning for a friendly unit in an offensive (attack) posture. The example has four available subordinate units, five V-Lanes 501, and six Task Organizable Subordinates 801C. From this input the METT-T Parser 105 populates the following FCOA Variables 107:
In practice, Bypass Criteria has a strong effect on the offensive tactic. A low Bypass Criterion is a conservative tactic that typically produces modest results with modest risk. A high Bypass Criterion is typically considered a high-risk/high-payoff tactic.
For Reserve Guidance, the four possible employment philosophies are Stay in Lane, Best Dent, Best Hole, and First Hole. The first three philosophies control whether and where a reserve unit commits after the Reserve Threshold has been met. The first two philosophies do not require a penetration of opposing defenses for commitment. The Best Dent philosophy directs the reserve unit to the defense location closest to penetration. The Best Hole philosophy directs the reserve unit to the most significant penetration, whereas the First Hole philosophy commits the selected reserve subordinate unit to the first penetration, regardless of whether the Reserve Threshold has been met. The METT-T Parser 105 provides a full set of instances for this variable.
As represented by the “Manual Input” shape of the IPB ECOA box 108 in the FIG. 1 flow chart, a user may develop ECOA's by hand-selecting constituent COA Variables. In limited performance testing, an experienced user was able to accomplish this task in well under five minutes. This compares favorably to current practices that involve hand drawing COA graphics on an acetate overlay or on a computerized map.
In select embodiments of the present invention, an intelligence planner now employs select embodiments of the present invention to produce a set of friendly defensive COA's that in reality become the IPB set of ECOA's 108 the planner will use. The basic departure of the Reverse-IPB process 109 from the FIG. 1 flow chart procedures is at 122, 123, where a user (intelligence planner) now selects a set of representative offensive COA's, rather than a single COA for execution. This Reverse-IPB procedure 109 also provides an intelligence planner the ability to identify all reasonable options available to an opposing force (e.g., an enemy commander). All the other advantages as described above would also accrue to this IPB ECOA set 108. Thus the game-theoretic analysis is more rigorous, in turn enabling development of a comprehensive set of FCOA's.
In select embodiments of the present invention, unlike the counterpart COA selection set for FCOA candidates 111, a user stabilizes the ECOA IPB set 110 for much of the game-theoretic analysis. In other words, select embodiments of the present invention reject a co-evolutionary paradigm, in which a late-generation FCOA may be vulnerable to an early-generation ECOA. Since early-generation ECOA's are often not represented in later generations, a user (commander) may be unaware of this vulnerability when relying upon a “co-evolutionary” analysis.
Therefore, in order to provide a standardized analysis of each considered FCOA, an FCOA Evaluator 115 provides an end-state estimate of a submitted FCOA for each ECOA in the ECOA IPB set 110. A user retains the ability to modify the ECOA IPB set 110 until ready to start systematic FCOA evaluation, at which time the ECOA IPB set 110 is “locked.”
In select embodiments of the present invention, it is sometimes preferable to modify the IPB ECOA set 108 after an initial analysis of the relative merits of two or more FCOA's has been completed. If a user later adds to or modifies the IPB ECOA set 108, all relevant changes to FCOA's should be re-submitted to the FCOA Evaluator 115 to insure a corresponding updated evaluation.
In select embodiments of the present invention, the standardization of the evaluation metric is guaranteed by “locking in” the IPB ECOA set 108 as well as the Desired End State evaluation criteria 114 at the FCOA Optimization thru the Genetic Algorithm (GA) 119. Locking in the IPB ECOA set 108 insures that all FCOA's considered by the GA use the same evaluation metric.
Select embodiments of the present invention allow for a secondary, co-evolutionary-like analysis at the FCOA Vulnerability Analysis 124. The FCOA Vulnerability Analysis 124 uses steps of the process recursively, such as the Reverse IPB 109 procedures, to develop ECOA's optimized against a chosen FCOA. An automated FCOA Vulnerability Analysis 124 allows a planner to gain a comprehensive set of Most Dangerous ECOA's and associated (Battle) Scripts. This significantly improves a unit's ability to prepare countermeasures for each vulnerability.
In select embodiments of the present invention, if an intelligence planner has established a truly representative IPB ECOA set 108, the FCOA Vulnerability Analysis 124 will yield a similar ECOA set. If the FCOA Vulnerability Analysis 124 does identify a new reasonable ECOA or ECOA set, then a user (commander) should consider re-initiating the entire planning process from the ECOA IPB set 110, adding the newly identified ECOA or ECOA set to the previous ECOA IPB set 110. Otherwise, a selected FCOA may be vulnerable.
In select embodiments of the present invention, the Terrain Informed War Game Model 112 conducts a fast-abstract simulation of combat for the selected COA's and outputs a time-phased estimate of the location and strength of all selected subordinate units during engagement (battle). The simulation is fast, typically on the order of 10 milliseconds, because the Game Board, Game Pieces, Game Strategies (tactics), and Game Rules are all highly abstracted to support fast calculations in RAM of estimates such as attrition.
In select embodiments of the present invention, time-phased snapshots of selected unit disposition and strength that are output from the Terrain Informed War Game Model 112 are realistic estimates of combat. This is because abstractions of all game objects are crafted to retain only that information pertinent to aggregate unit attrition and maneuver posture.
A realistic, 10-millisecond estimate of combat is a useful tool in the tactical planning domain. In limited performance testing of select embodiments of the present invention by domain experts, it is common for users to direct war gaming analysis of thousands of battles in a few minutes. Select embodiments of the present invention employ repeated, game-theoretic submissions of FCOA's and ECOA's to a Terrain Informed War Game Model 112. Since a battlefield is a chaotic system, context-appropriate tactics are not derivable from first principles. The speed of a Terrain Informed War Game Model 112, directed by a suitable optimization strategy such as the FCOA Optimization thru a Genetic Algorithm 119, facilitates development of an “emergent intelligence” on the appropriate tactic to use in a particular METT-T situation. In limited performance testing, this emergent intelligence is subjectively comparable to the insight of experienced domain experts.
The ability of select embodiments of the present invention to develop appropriate tactics in a given situation does not replace human experts. Rather, select embodiments of the present invention Cognitively Amplify the judgment of human experts to produce an FCOA analysis that is more rigorous, more comprehensive, and completed much faster than possible previously.
Refer to FIG. 9 listing component sub-processes within the Terrain Informed War Game Model 112 employed with select embodiments of the present invention. These include:
If a user executes a war game for the purpose of Visualization 113, the Terrain Informed War Game Model 112 outputs the entire set of battle snapshots to the visualization service. If the purpose of the war game is to support the FCOA Evaluator 115, the Terrain Informed War Game Model 112 outputs only the last snapshot of the engagement, since the FCOA Evaluator 115 focuses only on end state (final snapshot). This technique reduces CPU time and conserves computer memory, enabling evaluation of thousands of FCOA's per minute.
Calculate Attrition for Tokens (Units) in Contact 112D is a key aspect of the “Terrain Informed” feature of the Terrain Informed War Game Model 112. In establishing an engagement (e.g. a firefight), the Terrain Informed War Game Model 112 determines the relative combat power of each token (unit), as modified by the local terrain effects abstracted by the Braswell Index in that mobility corridor 301. In select embodiments of the present invention, this simulation rewards tokens (units) that intelligently leverage terrain, and appropriately punishes tokens (units) that disregard terrain characteristics. In select embodiments of the present invention, the Terrain Informed War Game Model 112 then consults an implementation of the well established Dupuy QJMA attrition model to determine how much attrition each token should suffer during subsequent time slices. Since the Terrain Informed War Game Model 112 assesses attrition during each of many time slices, the net effect is a discretized approximation of the widely-accepted Lanchester Differential Equation for combat attrition.
For Each Token (Unit), Assess Attrition and Update Status 112E: in select embodiments of the present invention after attrition has been assessed for all active participants, the Terrain Informed War Game Model 112 compares the status of each token (unit) with the criteria, thresholds, and policies in the executing COA, and takes appropriate action. For example, if a token has suffered attrition to below the Withdrawal Criteria specified for that token in the corresponding COA Variable, then the Terrain Informed War Game Model 112 withdraws that token from engagement (combat), and de-activates the engagement (e.g., afirefight) as appropriate. Withdrawn tokens are not eligible for any movement or other actions for the remainder of the engagement.
Snapshots are “taken” during the course of an engagement (e.g., a battle) and the Terrain Informed War Game Model 112 develops a time-phased set of these snapshots, one per time slice, that is used for later evaluation (FCOA Evaluator 115) or Visualization 113.
Engagement (Battle) Termination Criteria Test 112G is exercised by the Terrain Informed War Game Model 112. Examples of termination criteria may be: all tokens have withdrawn from engagement; all offensive tokens have withdrawn from engagement, and defensive actions have ended; most tokens have withdrawn and remaining tokens are in reserve and thresholds have not been exceeded (and logically can no longer be exceeded). In select embodiments of the present invention, if the Engagement (Battle) Termination Criteria have not been met, the Terrain Informed War Game Model 112 returns its processing flow 901 to the Increment Time Slice Counter 112B. This iterates another sequence of token (unit) movement, attrition, and status updates described in Sub-steps 112C through 112F. If the current engagement (battle) state passes the Engagement (Battle) Termination Criteria Test 112G, then the Terrain Informed War Game Model 112 exits the loop 901, and finalizes the Snapshot S et 112H for use by the FCOA Evaluator 115 and for Visualization 113.
In select embodiments of the present invention, Visualization 113 allows a user to gain a time-space appreciation of the engagement (battle) dynamics for the selected FCOA 111 and ECOA 110 IPB sets. This is particularly important if the FCOA candidate 111 is computer nominated, from the FCOA Optimization thru a Genetic Algorithm 119. The animation allows a user to quickly understand the strengths and weaknesses of computer-nominated FCOA candidates 111.
Refer to FIG. 10, a sample screen print 1000 of a screen that a user may employ to establish Total Evaluation Criteria 1001 for a Desired End State 114. In select embodiments of the present invention a user (commander) may use this screen to alter default criteria. In select embodiments of the present invention, major criteria categories for a user's (commander's) consideration may be:
In select embodiments of the present invention a user (commander) develops a Desired End State 114 by selecting a combination of these criteria that reflects how a user would like the game board (battlefield) to look at the end of a successful mission. A user (commander) may establish a weighting scheme to reflect relative preferences for each of the criteria.
Select embodiments of the present invention automatically establish a default set of criteria for a Desired End State 114 to support staff planning and analysis prior to a user's (commander's) formal establishment of criteria. The un-weighted default set may include optimization of overall friendly strength and minimization of overall enemy strength. If a friendly unit is in attack, the default set also may include criteria for maximizing attacker end strength at mobility corridors 301 that end each of the V-Lanes 501, roughly equating to optimizing forces on mission objectives. If a friendly unit is in defense, a default set includes a minimization of attacking end strength at the ends of the mobility corridors 301 of the V-Lanes 501.
Overall Unit Criteria 1002 form the basis of an “enemy based objective” in which attrition of an enemy force is the prime consideration.
Time Criteria Candidates 1005 allow a user to select how time, a critical consideration in accomplishment of a mission, affects performance. This variable allows a user (commander) to explicitly model the importance of time.
With Specific Unit Criteria 1004 a user (commander) may specify a particular token (unit) regardless of employment or may specify uncommitted reserves, regardless of which tokens (units) the COA assigns as a reserve.
In select embodiments of the present invention, the FCOA Evaluator 115 returns the result as the score for that evaluation criterion against that particular “evaluated” ECOA from the ECOA Candidates 110. Refer to FIG. 12, a sample screen print 1200 illustrating a screen representing the predicted performance of a first considered FCOA, FCOA-1, given selected evaluation criteria 1201 in which a first criterion (Maximize Overall Attacker Strength) delivers scores 1203 of approximately 55.0, 62.8, and 73.1 for each of three ECOA's (ECOA-1, ECOA-2, ECOA-3) 1202 in the ECOA IPB Set 110. A user (commander) now knows that by selecting FCOA-1 for battle while the enemy selects ECOA-1, the friendly (attacking) strength at the end of the engagement (battle) will likely be 55%.
In select embodiments of the present invention, the GA typically finds some FCOA solutions that may not have immediately occurred to the human expert and serves as a check on the human planning process. Also, the GA typically does a better job than the human expert in fine tuning obvious tactical concepts into higher-scored FCOA Candidate sets 111. However, the GA cannot optimize the intangibles that are not explicitly represented in the Terrain Informed War Game Model 112, such as the personalities of subordinate unit commanders and staffs. Thus a user should exercise judgment when reviewing scores from the FCOA Evaluation 116. In summary, for select embodiments of the present invention it is generally prudent to employ a combination of manual and GA optimization to develop a “best” FCOA Candidate set 111.
With select embodiments of the present invention, when a user initiates the Risk Deprecation Analysis 120, a computer re-evaluates each Results Matrix of the FCOA Evaluation 116 a number of times, “deprecating” a different ECOA Results column (FIG. 12) each time. If there are 100 candidate FCOA's in the FCOA Candidate set 111 and three ECOA's in the ECOA IPB set 110, this yields 300 deprecated result matrices as explained below. As an example, FCOA-54 will first be re-evaluated through a deprecation of ECOA-1 that has only the results of FCOA-54 vs. ECOA-2 and ECOA-3, dropping (deprecating) ECOA-1. Likewise, FCOA-54 will have two more “analysis” (deprecated) matrices from deprecating each of ECOA-2 and ECOA-3, respectively yielding a second deprecated FCOA-54 analysis matrix comparing ECOA-1 and ECOA-3 and a third deprecated FCOA-54 analysis matrix comparing ECOA-1 and ECOA-2. An articulated, automated Risk Deprecation 120 analysis enables a user (commander) to quickly understand the risk of each of the FCOA candidates 111 relative to any ECOA in the ECOA IPB set 110.
Refer to FIG. 14, a sample screen print 1400 of a page of a table summarizing a deprecated risk analysis as may be employed with select embodiments of the present invention. The first column in this sample 1400 displays the FCOA Name. In this example it is apparent that there are three rows of deprecated analyses for each individual FCOA. The second column, titled D-ECOA 1401, gives the ECOA that is deprecated for that row. The third column, title OR 1402, shows the Original Ranking (OR) of the un-deprecated analysis for that individual FCOA. The fourth column, titled DR 1403, shows the Deprecated Ranking (DR) reflecting the “merit” of that individual FCOA relative to the other candidate FCOA's when the deprecated ECOA is temporarily dropped (i.e., deprecated) from the ECOA IPB set 110. For example, the merit of CN-476 (Gen-2) in row 1 is 30 compared to others but its original ranking was 7. The fifth column, titled CR 1404, shows Change-In-Rank (CR) from the deprecated analysis. For the example of row 1 above, the change is 7-30 or -23. Another example demonstrates an advantage of a deprecated analysis. The candidate FCOA CN-367 (Gen-2) (rows 12-14) is ranked 12th when evaluated against the entire ECOA IPB set, 110, but moves up to first when the ECOA termed “Strong Right” (row 13) is deprecated. This indicates that CN-367 (Gen-2) is particularly vulnerable to the Strong-Right ECOA, but is otherwise an extraordinary FCOA. Note that select embodiments of the present invention provide for weighting both the DR and the CR as indicated in the columns designated DRW 1405 and CRW 1406, respectively. The sample 1400 also indicates an opportunity for a user (commander) to highlight 1407 values in a column to further facilitate the decision making process by choosing any of “greater than selection,” “lesser than selection,” or “no highlight.”
With select embodiments of the present invention, the Risk Deprecation Analysis 120, allows a user (commander) to quickly find and consider risks associated with the FCOA's that show deprecations of interest, like CN-367 (Gen-2). For example, an intelligence planner may estimate that a Strong Right ECOA is not very likely and a commander may decide to accept that risk. This risk may be more readily accepted if the intelligence planner assures the commander that intelligence is available to determine if the enemy has selected a Strong Right ECOA prior to having to engage a contingency (branch) plan optimized against the Strong Right ECOA. Evaluation Criteria Deprecation Analysis 121 (FIG. 1). With select embodiments of the present invention, a user (commander) has the option (recommended) of also conducting an Evaluation Criteria Deprecation Analysis 121, a counterpart of the Risk Deprecation Analysis 120. The Evaluation Criteria Deprecation Analysis 121 deprecates a user's (commander) Evaluation Criteria for the Desired End State 114, one at a time, rather than the ECOA's from the ECOA IPB set 110. Typically, a user (commander) may have established the original Evaluation Criteria for the Desired End State 114 without significant analysis of the tactical feasibility. Thus, the Evaluation Criteria Deprecation Analysis 121 enables a user (commander) to fully evaluate the “cost” of each evaluation criterion in terms of finding an FCOA that would otherwise score well against the remaining (non-deprecated) evaluation criteria. As a result of this second deprecation analysis, a user (commander) may decide to accept an FCOA with an otherwise low score, since the cost of a particular deprecated evaluation criteria is much more than originally anticipated, given a rigorous war gaming analysis of the tactical situation. In select embodiments of the present invention, the automation for the Evaluation Criteria Deprecation Analysis 121 parallels the Risk Deprecation Analysis 120 and the table (matrix) screen is similar to that of the sample 1400 in FIG. 14. An automated, articulated Evaluation Criteria Deprecation Analysis 121 allows a user (commander) to quickly understand the relative cost of each of the criteria used to evaluate the FCOA's for the original Desired End State 114.
In summary, select embodiments of the present invention are much faster and provide significantly more game-theoretic comprehensive analysis than existing approaches. In select embodiments of the present invention, the collective set of procedures (FIG. 1) provides Cognitive Amplification for MDMP and IPB planners, permitting better and faster analysis of FCOA's and ECOA's in significantly greater quantities than heretofore possible. Select embodiments of the present invention allow human experts to concentrate on conducting Military Art and a computer to execute Military Science at the direction of human users. Conventional computer reasoning processes are incapable of knowledge level interactions since the engagement context is only available in the minds of skilled planners, such as those military planners conducting MDMP and IPB. With select embodiments of the present invention, providing computer-based emergent intelligence quantifies many more reasoning processes beyond “fast and good” MDMP and IPB.
The abstract of the disclosure is provided to comply with the rules requiring an abstract that will allow a searcher to quickly ascertain the subject matter of the technical disclosure of any patent issued from this disclosure. (37 CFR §1.72(b)). Any advantages and benefits described may not apply to all embodiments of the invention.
While the invention has been described in terms of some of its embodiments, those skilled in the art will recognize that the invention can be practiced with modifications within the spirit and scope of the appended claims. For example, although the system is described in specific examples for decision support to military battalions, brigades, and division staffs, for both collective training and real operations, it may be also employed in such diverse applications as military institutional training of personnel on the MDMP process and Military Art; informal assessments of individual and staff capabilities in both Military Art and Military Science, similar to the U.S. Navy's Top Gun program and the U.S. Army's Combat Training Centers for entire units or the Ender's Game novel; decision support to military small unit leaders; establishing topographical mapping requirements for contingency support; support to operations research and systems analysis on tactical command and control systems; and commercial computer and video games both for recreation and for training.
In the claims, means-plus-function clauses are intended to cover the structures described herein as performing the recited function and not only structural equivalents, but also equivalent structures. Thus, although a nail and a screw may not be structural equivalents in that a nail employs a cylindrical surface to secure wooden parts together, whereas a screw employs a helical surface, in the environment of fastening wooden parts, a nail and a screw may be equivalent structures. Thus, it is intended that all matter contained in the foregoing description or shown in the accompanying drawings shall be interpreted as illustrative rather than limiting, and the invention should be defined only in accordance with the following claims and their equivalents.
A Genetic Algorithm (GA) is a well-understood optimization technique for problem solving, as described by David E. Goldberg. Goldberg, David E., Genetic Algorithms in Search, Optimization, &Machine Learning, Addison-Wesley Publishing Company, Inc., Reading, Mass., ISBN 0-201-15767-5, 1989. (Goldberg). Genetic Algorithms emulate the mechanics of natural selection and natural genetics by “breeding” better solutions to a problem from the genetic material of previous solutions. As described on page 33 of Goldberg's text, the foundation for GA theory is governed by John Holland's “schema theorem” that relates the expected effects on a population from the manipulation of basic genetic operations: selection, cross-over, and mutation.
Select embodiments of the present invention implement a “Simple GA” (SGA) as described in Goldberg. This SGA represents all COA solutions, both FCOA's and ECOA's, as “bit strings” which the genetic operators can manipulate. A bit string is a structured sequence of binary bits (“0” or “1”). This implementation structures the bit string so that each COA variable is represented by a component bit string, mapped to a known position on the bit string representing the total COA solution. For example, a COA variable like “offensive reserve guidance” has four possible instances, and is represented by a two-bit segment, where “00” maps to the first possible instance (stay in lane), “01” maps to the second possible instance (best dent), and so on. A COA variable like unit assignments might have 120 possible instances, so it would be represented by a seven-bit segment, where “0000000” maps to the first possible instance, and the last (120th) possible instance maps to “1110111” (“119” in decimal, which is the 120th solution when counting starts at “0”). The remaining eight possible numbers in a seven-bit string (“1111000” through “1111111”) are ignored by this implementation, since they do not map to solutions. This implementation retains a “genetic reference map” of where bit strings are located on the total solution bit string, so that any bit string of the proper length can be de-referenced as a viable FCOA or ECOA for use with the Terrain Informed War Game Model 112.
Like any SGA described by Goldberg, this implementation breeds potential new solutions from bit strings by executing genetic operators. “One-point Crossover” is a basic example of a genetic operator, where two “parent” solutions “0000000” and “1111111” are crossed at the third position to create two “children” solutions of “0001111” and “1110000.” Early in the breeding program, one of the children is likely to produce a superior solution to both parents, and one of the children is likely to be inferior to both parents. To retain a constant-sized breeding pool, two solutions of this nuclear family are retained for future generations, and two are discarded. For a properly structured optimization problem (like selected embodiments of the present invention), the schema theorem guarantees the mathematical expectation that “good” schemas will increase in successive generation. This roughly translates into the notion that if a solution has a good “idea” of a particular variable instance, or combination of variable instances, that good idea can be expected to at least survive, and probably thrive in successive generations. Performance testing of select embodiments of the present invention has repeatedly validated this proposition, which means the SGA within this invention is likely to find very good FCOA's and ECOA's, if very good COA's can be reasonably found in a battle context.
Of course, this begs the question of what constitutes a good solution or COA. Select embodiments of the present invention enable a user to specify a desired end state for a battle situation. This desired end state is the commander's preferred disposition (location on the BBE Game Board) as relates the strength of both friendly and enemy forces at the end of a battle. A user may choose and weight a set of evaluation criteria unique to a particular situation. The categories of evaluation criteria are:
Maximize/Minimize Battle Time, in which the length of the battle is measured as a ratio against an exemplary long battle. An FCOA solution that produces long battles, relative to the exemplary long battle, against the ECOA IPB set 110 contributes to a higher FCOA score if a user has selected to maximize battle time. If a user has chosen to minimize battle time the contribution to the total FCOA score is 1.0 minus the ratio of battle time to the exemplary length battle.
Overall Unit End Strength provides for a user selecting either the total enemy unit or the total friendly unit for maximization or minimization. Select embodiments of the current invention measure the ratio of the designated unit strength at the end of battle to its strength at the beginning of the battle. If a user wishes to maximize that end state, then the contribution is the ratio (multiplied by 100 for normalization, as are all ratios in this part of the protocols). If a user desires to minimize the end state then the contribution is 1—the ratio (multiplied by 100).
Specific Unit End Strength allows a user to select a particular enemy or friendly unit for minimization or maximization, using the same start/end strength ratio described above.
Uncommitted Reserve End Strength allows a user to calculate the strength of all friendly (or enemy) units that are uncommitted at the end of battle and then divided by the start strength of the total set of friendly (or enemy) units to develop a customized ratio for a user that wants to maximize (or minimize) the amount of forces that are immediately ready for a follow-on battle.
Mobility Corridor End Strength allows a user to select a set of mobility corridors 301 on the BBE Game Board 403 and also to minimize (or maximize) enemy (or friendly) forces in that mobility corridor 301 at the end of battle. The end strength of the selected opponent in that mobility corridor 301 is calculated as a ratio of the start strength of the entire opponent start strength. This enables a user to specify mobility corridors 301 to secure (retain possession of) or deny to the enemy.
A user weights each of the selected evaluation criteria and the total set of evaluation protocols is considered in each battle for the FCOA evaluation. The resultant evaluation includes comparing an FCOA against each of the ECOA's in the ECOA IPB set 110. Further, a user may weight those ECOA's according to their relative importance or probability of adoption in the total evaluation. The FCOA Evaluator 115 then calculates an FCOA Evaluation 116 matrix for the evaluated solution (FCOA). This evaluation includes a single score that can roughly be interpreted as the “goodness” of the FCOA, with respect to a user's Desired End State 114, for the given METT-T battle situation. To ensure that scores of all FCOA's may be compared, select embodiments of the present invention “lock” the evaluation criteria and ECOA IPB set 110 before running the GA. This avoids the distraction of “co-evolution” in which ECOA's are allowed to evolve against evolving FCOA's. Co-evolution makes comparison of FCOA's extremely difficult because of inconsistent evaluation standards.
The integration of evaluation criteria with the GA makes select embodiments of the present invention adaptable to the METT-T battle context as well as enabling a user to change evaluation criteria to explore options. If a user decides to value the left side of the battlefield more than the right side, the corresponding mobility corridors 301 are selected as evaluation criteria, and the GA breeds FCOA's that are more likely to satisfy the modified criteria. As an additional example, if a commander decides to destroy a particular enemy unit he can manipulate the evaluation criteria so as to promote the GA to breed FCOA's that are more likely to produce that result.
Military doctrine such as Army Field Manual 34-130, Intelligence Preparation of the Battlefield, 1994, strongly recommends detailed terrain analysis prior to the development of COA's. The U.S. Army Corps of Engineers' Topographic Engineering Center has developed a powerful suite of semi-automated terrain analysis tools to support this task, but the resulting products are extremely large, typically measuring tens or hundreds of megabytes. Battle simulations that use these products generally produce more realistic estimates of combat attrition but those simulations run extremely slow relative to select embodiments of the present invention. This is due to the need for the simulation to continually access terrain analysis data on a computer's hard drive rather than in RAM.
Refer to FIG, 3. The Braswell Index retains pertinent effects of the terrain analysis by grouping them according to mobility corridors 301 between obstacles (to Unit Maneuver) 302. On a map those mobility corridors 301 are represented by polylines 301. Since units can maneuver around obstacles 302 and through mobility corridors 301, the representative polylines 301 form a network topology that may be used to support maneuver analysis for tactical units. This maneuver network forms an infrastructure for the BBE Game Board 403 (FIG. 4) that calculates unit maneuver via the polylines 301, although the real-world representation of that abstraction is a unit that maneuvers through the represented mobility corridor 301 represented by the polylines 301 as the “center” of the mobility corridor 301. The Braswell Index further empowers the BBE Game Board 403 by characterizing the combat effects of terrain in each two-dimensional (2D) mobility corridor 301 along each representative one dimensional (1D) polyline 301. For example, if the cross country mobility (CCM) characteristics of the terrain within a mobility corridor 301 significantly favor the defense, then the Braswell Index records an appropriately high combat multiplier for the BBE to use in calculating attrition for any simulated combat in that mobility corridor 301. The defender's combat strength will correspondingly increase due to having appropriately leveraged the characteristics of terrain. If combat moves to another mobility corridor 301 that favors the attacker rather than the defender select embodiments of the invention may calculate attrition with an appropriate multiplier for the attacker rather than the defender.
The Braswell Index calculates these terrain effects for the major elements of a tactical terrain analysis and records multipliers for both attacker and defender in the reference for each mobility corridor 301. Because the Braswell Index stores the combat effects as simple multipliers in each mobility corridor 301 rather than the associated geo-spatially rectified terrain analysis product, the size of the BBE Game Board 403 is typically several thousand times less than traditional terrain analysis products—kilobytes compared to megabytes. As a result, the Terrain Informed War Game Model 112 loads the BBE Game Board 403 into RAM (fast) memory (microsecond access) rather than having to rely upon a slow hard drive (millisecond access). This is why the battle simulations in select embodiments of the present invention estimate combat attrition several thousand times faster than other combat attrition models and still retain realistic modeling of the terrain's substantial influence (effects) on force attrition.
Refer to FIGS. 5A and 5B. The Braswell Index forms the basis for the BBE Game Board 403, depicting where units can maneuver. However, military units typically do not maneuver independently in combat. They typically maneuver as part of a formation of cooperating units. The movement of these unit formations requires more terrain analysis than the Braswell Index provides. This additional analysis is provided in the Articulated MCOO 101 produced by the MCOO-Maker application developed by the Topographic Engineering Center. The Modified Combined Obstacles Overlay (MCOO) is a traditional Army IPB product (FM 34-130) that enables a planner to understand the total unit maneuver formations supported by the terrain. Military units typically attack from a line of departure (LD), which can be considered the start line for attack, and they maneuver towards a finish line, which is usually called the “objective.” The Braswell Index only shows options for maneuver, whereas the MCOO displays the routes of connected mobility corridors 301 from the anticipated starting line or area (LD) to an anticipated end point or area (objective). These routes are represented by associated polylines 301 as shown in FIG. 5A.
Conventionally, cooperating military units do not cross paths in an attack since that would significantly increase the risk of fratricide (friendly fire casualties), with almost no tactical advantage. Therefore, the purpose of an MCOO is to show how a formation of units may traverse an area of operation (AO) without crossing paths. This infers that network flow algorithms are more appropriate than normal path finding algorithms. However, an attacking unit “flows across” an AO only once rather than continuously. This merits an adjustment of the analysis from a network flow algorithm to a “network pulse” calculation. The MCOO-Maker conducts this network pulse calculation on the Braswell Index, and considers how many units can conceivably attack abreast in each mobility corridor 301. In FIG. 5A three units can attack abreast in the mobility corridor 301 that converges the three routes. The MCOO-Maker further develops logically-parallel routes using the V-lanes 501 that do not cross paths, but may share the same mobility corridor 301 on occasion. In those events, the mobility corridor 301 is virtually divided into sub-mobility corridors to support multiple units maneuvering through the same mobility corridor 301. The MCOO-Maker further refines this Articulated MCOO 101 by separating the logical paths into geo-spatially rectified paths of connected mobility corridors termed Virtual (V) Lanes 501 that form a combat maneuver matrix for the BBE to employ.
In select embodiments of the present invention, V-Lanes 501 form logical maneuver options for an attacking unit whereas the Lines of Defensible Terrain (LDT's) 401 FIG. 4 form employment options for a defending unit. If the essence of the attack is to deliver an effective network pulse, then the essence of the defense is to deploy an effective “network block” that prevents any of the attacking pulse from reaching the objective line. The LDT's 401 are a cooperating set of mobility corridors 301 that effectively form a network block of the MCOO's network-pulse attack analysis. As a result, the Articulated MCOO is the meaningful addition to the Braswell Index of V-Lanes 501 (for formation attack analysis) and LDT's 401 (for formation defense analysis). Together, the Articulated MCOO and the Braswell Index form the basic Game Board 403 for the BBE war games.
The Mission, Enemy, Terrain, Troops, and Time (METT-7) Parser 105 employed with select embodiments of the present invention evaluates battlefield options in the form of COA variables for both attackers and defenders. A Course of Action (COA) may be either an offensive or defensive COA; assigned to an enemy as an Enemy COA (ECOA), or to a Friendly as a Friendly COA (FCOA). As a set, the FCOA variables for a situation define the major options for a friendly commander in developing an Operations Plan (OPLAN) that directs a unit in combat. COA variables are discussed in the Detailed Description above.
COA Variables are highly dependent upon the METT-T situation. For example, if an area of operation (AO) is small a commander has limited options and if the force comprises three subordinate units instead of five, options are further constrained. The METT-T Parser 105 conducts an inventory of the terrain Game Board 403 by employing the Articulated MCOO and Braswell Index 101. The more V-Lanes 501 available in an Articulated MCOO the more options a friendly commander has to consider. Several LDT's 401 provide more options for a defending unit whether friendly or enemy. The METT-T parser 105 conducts an inventory of the game pieces (units) by examining the Friendly OB 103 and Enemy OB 102. The METT-T Parser 105 then consults the Missions and Postures 104 identified by a user. This indicates which side is attacking and which side is defending.
The METT-T Parser 105 then develops a master FCOA Variable Set 107, as well as a master ECOA Variable Set 106. These sets include all possible instances for each COA Variable in each set. Refer TO FIG. 6A. For example, if there are seven V-Lanes 501 in an area of operations, and a commander desires to attack two units abreast, this automatically implies one subordinate unit boundary between the two lead subordinates. The METT-T Parser 105 develops and maintains a set of options for this COA Variable for all possible locations of that one subordinate unit boundary. In this example there are six possibilities, one each between each pair of neighboring V-Lanes 501. If the METT-T Parser 105 receives nine V-Lanes 501 from the Articulated MCOO, then the METT-T Parser 105 develops eight possible instances for that COA Variable. In the same manner, the METT-T Parser 105 develops all possible instances for all other COA variables in the battle situation.
By developing and maintaining all possible variable instances, select embodiments of the present invention enable a user to model any possible COA for selection into a ECOA IPB Set 110 and the FCOA Candidates Set 111. The Terrain Informed War Game Model 112 then provides a user the ability to understand tactical consequences of selected COA's.
Refer to FIG. 1. The Terrain Informed War Game Model 112 is a fast, abstract battle simulator that produces realistic estimates on final disposition and strength of opposing units, given a select FCOA from the FCOA Candidates set 111 and a select ECOA from the ECOA IPB Set 110. This battle simulator is abstract in that it represents only a small percentage of terrain and force information available in conventional training simulations. It is a realistic simulation because the information it maintains and uses is only that most pertinent to calculating attrition and maneuver estimates. The resultant simulation is fast compared to conventional simulations, e.g., a battle is simulated in a few milliseconds vice a few hours. As implemented in select embodiments of the present invention, the Terrain Informed War Game Model 112 provides cognitive amplification for a user, enabling development and analysis of a far greater number of COA's than otherwise possible.
The Terrain Informed War Game Model 112 provides this cognitive amplification by working together with other elements of select embodiments of the present invention. A user provides Mission, Enemy, Terrain, Troops, and Time (METT-T) battle context to this game “engine” by establishing a game board 403 from the Articulated MCOO and Braswell Index 101, as well as game pieces (tokens) from the Enemy Order of Battle 102 and the Friendly Order of Battle 103. A user also provides a purpose to the war game through a Missions and Postures input 104. The METT-T Parser 105 refines these inputs into a form usable by the Terrain Informed War Game Model 112 by developing the ECOA Variable set 106, FCOA Variable set and the METT-T context (or battle situation) for the Desired End State interface 114. The ECOA Variable Set 106 enables a user to develop an ECOA IPB set 110 to facilitate analysis of various scenarios. The ECOA IPB set 110 represents an opposing force's (enemy commander's) major options in an upcoming engagement (battle, firefight, and the like). A user may also employ elements of select embodiments of the present invention to conduct a Reverse IPB process 109 to provide a reasonable survey of enemy tactical options. The FCOA Variable Set 107 enables a user, or the Genetic Algorithm 119 to develop a set of FCOA's that provide options (FCOA's) to employ.
A user has some pre-conceived notions about the relative merit of various FCOA's and ECOA's being considered but these fall short of a comprehensive analysis in a game-theoretic context. The Terrain Informed War Game Model 112 assists a user in conducting a comprehensive analysis by developing a reasoned estimate of the likely outcome of any selected FCOA against any selected ECOA. Since the ECOA Variable Set 106 and FCOA Variable Set 107 support the development of any and all COA's, within the limits of a combat model, the Terrain Informed War Game Model 112 examines any possible combat interaction in a particular METT-T circumstance. A user may employ the Terrain Informed War Game Model 112 to verify any pre-conceived notions about the strengths of particular FCOA's against select ECOA's. A user then “fine tunes” preferences through manual FCOA Optimization 118. Alternatively, or in addition to, a user may employ the Genetic Algorithm (GA) 119 to automatically survey the tactical METT-T context and determine which FCOA's will succeed against the ECOA IPB Set 110.
As will be described later in this appendix, the Terrain Informed War Game Model 112 maneuvers units across the BBE Game Board 403 in accordance with the governing COA for that unit, i.e., the “selected” FCOA for friendly tokens and the “selected” ECOA for enemy tokens. When opposing tokens engage within the same mobility corridor 301, the Terrain Informed War Game Model 112 estimates the duration of the ensuing engagement, as well as the likely attrition probability for each token, given the current strength of the units, their posture, and the corresponding tactical effects of the local terrain, the latter being a reference “call” to the appropriate Braswell Index multiplier for that mobility corridor 301. After each discrete “clock tick” in the simulation, the Terrain Informed War Game Model 112 checks conditional rules found in the governing FCOA and ECOA. These rules have thresholds and other conditions that may trigger a withdrawal or a call for local reinforcement by a reserve unit.
Depending upon the governing COA, units may stay in local engagement for several clock ticks before a terminating event like withdrawal, bypass, or annihilation. If a defender has chosen an appropriate defensive COA against an attacker's offensive COA, then no attacking units may reach the objective. If the defensive COA is mis-matched against the offensive COA, then some or all of the attacking units may reach the objective.
The Terrain Informed War Game Model 112 terminates the engagement when all units on both sides have finished their maneuvers and “contacts.” The Terrain Informed War Game Model 112 then delivers “(battle) snapshots” to other elements of select embodiments of the present invention. A snapshot is a simple report for every token of its percentage strength and disposition (location) with respect to the respective mobility corridors 301. If the Terrain Informed War Game Model 112 is reporting to the (Battle) Visualization service 113, then it sends a snapshot for every clock tick of the simulated engagement for subsequent animation and visualization. If the Terrain Informed War Game Model 112 is reporting to the FCOA Evaluator 115, then it only sends a snapshot for the final clock tick of the engagement for comparison to the Desired End State 114.
As described in Appendix A the Genetic Algorithm employed with select embodiments of the present invention allows a user to specify evaluation criteria for the Desired End State 114. The FCOA Evaluator 115 uses these evaluation criteria to provide an objective, normalized score for each engagement. If the score is high, then the FCOA accomplished most of the criteria established in the Desired End State 114 against the ECOA selected for that engagement. However, an enemy may not select that particular ECOA in actual battle, so military IPB Doctrine (e.g., Army doctrine specified in FM 34-130) recommends that a user war game an FCOA against all options represented by the ECOA IPB Set 110. The FCOA Evaluator 115 automatically engages the Terrain Informed War Game Model 112 to produce each score for the “evaluated” FCOA against each of the ECOA's in the ECOA IPB set 110. This results in a matrix, an example of which is shown in FIG. 12. The cumulative score 1204 for the FCOA represents an objective assessment of that FCOA's goodness as measured against all evaluation criteria and against all war gamed ECOA's. As described in Appendix A the Genetic Algorithm uses this cumulative score 1204 as fitness criteria when searching for better FCOA's.
Refer to FIG. 9. The section above explained how certain elements of select embodiments of the present invention employ the Terrain Informed War Game Model 112. The following explains the internal workings of Terrain Informed War Game Model 112 in more detail. The Terrain Informed War Game Model 112 initializes the METT-T context of a battle by importing the Articulated MCOO and Braswell Index 101, the Enemy OB 102, the Friendly OB 103, a selection from the ECOA IPB set 110, and a selection from the FCOA Candidates 111. As displayed at 112A, the Terrain Informed War Game Model 112 then consults the two governing COA's to array initial game pieces at their starting locations. As an example, a Defensive COA, either the ECOA or FCOA, specifies the Line of Defensible Terrain (LDT) 401 along with the specification of which forces should be arrayed along that LDT 401 and where.
The Terrain Informed War Game Model 112 then initiates a “maneuver-attrit-react” cycle for each clock tick in the simulated engagement. When a user establishes the Missions and Posture 104 for the simulation, a “time slice” or clock tick duration is established that may vary from six minutes to 30 minutes. Thus, an engagement that lasts three hours has 30 clock ticks if the time slice is set to six minutes. If the time slice is set to 30 minutes that engagement has only six clock ticks. The maneuver-attrit-react cycle displayed in FIG. 9 is executed once for each clock tick.
At step 112B the Terrain Informed War Game Model 112 increments the clock tick to indicate another time slice should be simulated. At step 112C the Terrain Informed War Game Model 112 checks the status of each token with respect to the governing COA's directions. If the token is not engaged and the governing COA directs forward movement along a V-Lane 501 (FIG. 5), then the Terrain Informed War Game Model 112 calculates a forward displacement for that token consistent with its movement speed and the ability of the mobility corridor 301 to facilitate movement. Prior to reaching the calculated displacement location, the token may come within engagement range of an opposing token. In that event, the token stops and engages with the opposing token. Every token continues in this manner for step 112C of every clock tick in accordance with the directions of the governing COA.
At step 112D the Terrain Informed War Game Model 112 assesses attrition for all tokens engaged that clock tick. The amount of attrition is a function of the relative strengths of the opposing tokens and also the effect of the local terrain within the mobility corridor 301 to support the appropriate operations. The “current strength” property of each token is adjusted downwards appropriate to the attrition inflicted by the opposing token. In select embodiments of the present invention, this attrition model is derived from the Dupuy QJMA methodology, as explained in the Detailed Description. Since attrition is assessed every clock cycle, the result is a discrete approximation of the Lanchester Differential Equations, dA/dt=kD and dD/dt=k′A, where A is the strength of the attacker and D is the defender's strength. (Dupuy 1979, page 148). Some approximation of the Lanchester Differential Equations is preferred by most operations analysts in the community.
At step 112E the Terrain Informed War Game Model 112 compares the status of each token with the directions in the governing COA. As an example, both offensive and defensive COA's specify withdrawal thresholds for each subordinate unit. When the subordinate unit's strength is reduced below a threshold specified in the governing COA, the Terrain Informed War Game Model 112 changes its posture to withdrawn, marking it as ineligible for any engagement. Similarly, there is a threshold for the commitment of a reserve along with policy guidance for how to employ that reserve when committed. Depending upon the directions of the governing COA, the Terrain Informed War Game Model 112 may change an activated reserve outside its current V-Lane 501.
At step 112F the Terrain Informed War Game Model 112 creates a snapshot of the current game state. This snapshot is a record of each token's current mobility corridor 301, the progress through the mobility corridor 301, and its strength. If the simulation is to support later visualization, this snapshot is attached at the end of a set of snapshots for possible animation on a geospatial information system (GIS). If the simulation is for evaluation purposes only the last snapshot of the engagement is archived for later submission to the FCOA Evaluator 115.
At step 112G the Terrain Informed War Game Model 112 checks to see if the engagement termination criteria are met. If any token is still engaged the engagement continues and the simulation cycles control 910 back to step 112B, Increment Time Slice Counter. If all tokens are withdrawn or are at their final destinations (objective line), or are reserves that will never be committed (having not met their thresholds), then the engagement may terminate and the appropriate snapshot records returned to either the Visualization service 113 or the FCOA Evaluator 115.
Select embodiments of the present invention produce sets of FCOA's for a commander's consideration. The Commander's FCOA Decision 122 is the point within the Military Decision Making Process (MDMP) where the commander selects one FCOA 123 from the FCOA Candidates 111 for implementation as the unit's Operation Plan (OPlan). Elements of select embodiments of the present invention provide three tools to assist a commander in determining the best FCOA: a Risk Deprecation Analysis 120, an Evaluation Deprecation Analysis 121, and a Pareto Analysis (not shown in FIG. 1 but referenced as a “button” 1601 in FIG. 16).
The FCOA Evaluator 115 fires the evaluation protocols established by a user in the Desired End State 114 to acquire an objective score of a selected FCOA's relative merit against a selected ECOA. To fully evaluate an FCOA, the FCOA Evaluator 115 directs a full evaluation as illustrated in FIG. 11. The feedback arrow 1101 shows how the FCOA Evaluator 115 initiates a simulation between an evaluated FCOA and each of the ECOA's in the ECOA IPB Set 110. This more comprehensive evaluation of the FCOA results in a cumulative evaluation score for the evaluated FCOA.
Consider a situation where there are two FCOA's being evaluated against an IPB set of three ECOA's, both FCOA's having a total evaluation score of 200. At first glance, these FCOA's seem equivalent but upon further examination of this surprisingly typical case the first FCOA, “FCOA-Home-Run” achieves a cumulative result by scoring 100 normalized points against ECOA-1, 95 points against ECOA-2, and only five points against ECOA-3. In contrast, the other FCOA option, “FCOA-Base-On-Balls,” achieves a cumulative result by scoring a more even distribution of 66, 67, and 67 points against the respective ECOA's. Clearly the two FCOA's are not equivalent in spite of having the same cumulative score of 200. FCOA-HR does extremely well against the first two enemy options but is annihilated by the third. FCOA-BB does medium well against the entire set of ECOA's.
The cumulative score hides this clear distinction between the two FCOA's, a distinction that a commander would want to consider before making a decision, since the HR option provides high-risk for high-gain, whereas the BB option provides low risk for medium gain. Before making a decision, a typical commander would discuss with his intelligence officer the probability that the enemy would employ its third option (ECOA-3), and the likelihood that re-directed reconnaissance assets could verify or deny enemy employment of that option in a timely enough manner to switch game plans (from HR to BB). Unfortunately, the commander and intelligence officer might not think to have this discussion if they only consider the false equivalent cumulative score of 200 for both FCOA's. The Risk Deprecation Analysis tool 120 provides the commander the analysis required to discover these false equivalencies and to quickly understand the dynamic relationships between each FCOA and each of the ECOA's in the IPB set.
FIG. 14 shows the results of a Risk Deprecation Analysis 120 in cooperation with other elements of select embodiments of the present invention. The Terrain Informed War Game Model 112 has conducted simulations between every FCOA candidate 111 and every ECOA in the IPB set. In this case there are 420 FCOAs in consideration, and three ECOA's (named “Strong Right,” “Strong Left,” and “Balanced”). Thus, there have been 1260 simulations and evaluations (420×3). FIG. 14 shows the details of some of the 1260 evaluations but these evaluation details are somewhat counter-intuitive and require explanation.
The first column gives the FCOA's name, all of which are “computer nominated” (CN) in this example. Since there are three ECOA's in the IPB set, each FCOA undergoes deprecation analysis three times, thus each FCOA is listed in three rows. The second column 1401 lists the name of the deprecated ECOA for that line's analysis. In the example of the first row the Strong Right ECOA has been deprecated, meaning that row shows the cumulative results of FCOA-CN-476 against the other two ECOA's, but not the deprecated ECOA-Strong Right. The third column 1402 shows the original, non-deprecated ranking of that FCOA, which in the case of the first row is 7th out of 420. The fourth column 1402 shows the new, deprecated ranking of the FCOA, which in the case of the first row is now 30th out of 420. In other words, when ECOA-Strong Right is thrown out of the IPB set, FCOA-CN-476 does not do as well as it did with the non-deprecated ECOA. The fifth column 1404 shows a change of ranking of −23, meaning the FCOA (CN-476) dropped from 7th to 30th. The tactical meaning of this analysis of the first row is that CN-476 does extremely well against ECOA-Strong Right, but not so well against the other two ECOA's. Columns 1405 and 1406 show the Deprecated Roulette Wheel and Change in Roulette Wheel rankings used to calculate the ordinal rankings in the previous columns.
In select embodiments of the present invention, a user does not need to study each of the 1420 lines in this table. When a user clicks on the column headings 1401 through 1406, the table re-orders itself in either ascending or descending order by that column's value. In other words, by clicking on the Change in Rank (CR) 1404 heading, the deprecated FCOA that benefited the most by dropping out an ECOA from evaluation will now be at the top of the table. If a user clicks again on the CR 1404 heading, the table re-orders itself in descending CR value, meaning the deprecated FCOA that was hurt the most by a missing ECOA is now at the top. When a user employs this re-ordering tool along with the greater-than and less-than highlighting tool 1407 at the bottom of the window, FCOA's with “interesting” sensitivities that are sensitive to specific ECOA's are quickly located. In limited performance testing of select embodiments of the present invention experienced users found extremely interesting FCOA-ECOA dynamics in less than one minute. This provides for a more informed command decision on which FCOA is appropriate for a mission.
The Evaluation Criteria Deprecation Analysis 121 works in a similar manner, but instead of deprecating ECOA's from the total evaluation, the Evaluation Criteria themselves are deprecated, one at a time. FIG. 12 shows an evaluation matrix for a typical FCOA, where the ECOA's are listed in the middle columns 1202, and the evaluation criteria are listed in the middle rows. The Risk Deprecation analysis 120 re-computes that matrix by dropping the middle columns 1202 (ECOA's) one at a time, whereas the Evaluation Criteria analysis 121 re-computes that matrix by dropping the middle rows (Evaluation Criteria) one at a time. The Risk Deprecation Analysis 120 quickly gives a user an idea of each FCOA's sensitivity to each ECOA, whereas the Evaluation Criteria Deprecation Analysis 121 quickly gives a user an idea of each FCOA's sensitivity to each Evaluation Criteria. For example, an Evaluation Criteria Deprecation Analysis 121 of FCOA-1 may show that FCOA-1 is unusually sensitive to Maximize Atk Strength at sub-MC 773, and that FCOA-1 would be an extraordinary solution if it were not for that one Evaluation Criterion. The commander may decide to trade off that Evaluation Criterion for superior expected performance in the other criteria. However, most commanders would want to consider that option, if it were available quickly and simply, which is the result of employing select embodiments of the present invention.
A third FCOA analysis tool, not displayed in FIG. 1 but inferred in FIG. 16 at 1601, is Pareto Analysis, a more “formal” analysis of the trade offs involved in FCOA evaluations. Consider the comparatively simple example of buying a car. In advance a buyer might state that she prefers quality over price and is willing to pay extra for a superior product. However, there might be a car that independent consumer agencies rank as 99% as good as another car, but at only half the cost. If a salesman (or a software cognitive amplification agent) consistently applies her a priori statement about preferring quality over price, then he (or it) would not present that second car as an option. But, common sense suggests that the buyer would certainly want to know about such an intriguing option.
A Pareto Analysis examines a set of solutions (cars, in the above example) by a relative comparison of each evaluation criteria (price and quality, in the above example), and eliminates Pareto-dominated solutions from consideration, in favor of Pareto-dominating solutions. In the car example, a third car might be better than the second car in both price and quality, in which case the second car is completely dropped from consideration since a rational consumer would consider the product in every evaluation criteria.
In this trivial example, a hundred cars could be Pareto-analyzed to find the two cars that together Pareto-dominate the other 98 but do not Pareto-dominate each other. One car is superior to the other in price and the other is superior in quality. The simplification of the large set of 100 cars to the small set of two cars is called the Pareto-optimal front, enabling a consumer to make a much simpler decision in confidence that she is getting one of the best possible cars out of 100 models, given her personal preferences.
Select embodiments of the present invention implement a Pareto-Analysis in a manner similar to the car example described above, but an increased number of evaluation criteria significantly increases the size of the Pareto optimal front. In limited performance testing of select embodiments of the present invention, an initial FCOA set of 420 solutions eliminates about 150 Pareto-dominated FCOA's from consideration, leaving about 270 remaining FCOA solutions on the Pareto-optimal front. Select embodiments of the present invention extend this multi-dimensional Pareto analysis (one dimension for each evaluation criteria). This Pareto functionality enables a user to determine the Pareto displacement of any Pareto dimension (evaluation functionality) between any two FCOA's on the Pareto optimal front.
Although this pair wise evaluation technique is difficult to use in a comprehensive manner against a large set of FCOA's, it is excellent for understanding the evaluation trade offs between any two FCOA's from a small set. In other words, after using the simple deprecation analysis tools to reduce the FCOA's in consideration from several hundred to just a handful, a user may employ the Pareto analysis tool to find interesting trade offs between FCOA's with respect to evaluation criteria. As an example, a user might discover that between two “finalist” FCOA's, one of the FCOA's is massively better in three evaluation criteria, whereas the second finalist FCOA is only moderately better in the remaining five evaluation criteria.
The net effect of these three tools enables users to quickly gain a sophisticated understanding of the relative advantages and disadvantages of a large number of FCOA's with respect to a large number of evaluation criteria and a large number of IPB ECOA's.