Agent-Oriented Architecture for Air Combat Simulation - CiteSeerX

Agent-Oriented Architecture for Air Combat Simulation April, 1993

Technical Note 42

By: Anand Rao Andrew Lucas David Morley Australian Arti cial Intelligence Institute Mario Selvestrel Graeme Murray Defence Science and Technology Organisation Aeronautical Research Laboratory

This research was supported by the Defence Science and Technology Organisation { Aeronautical Research Laboratory, a Generic Industry Research and Development Grant from the Department of Industry, Technology and Commerce, Australia and by the Co-operative Research Centre for Intelligent Decision Systems. Abridged versions of this paper appear in the Proceedings of the Australian Joint Conference on Arti cial Intelligence (AI-92), Hobart, Australia, 1992 and Proceedings of Future Directions in Simulation Systems Workshop, Melbourne, Australia, 1992.

Abstract

Air combat modelling using graphical simulation is a powerful means for development and evaluation of tactics. However, large models are particularly expensive and time-consuming to maintain and modify. Multi-aircraft full mission man-in-the-loop simulators will provide an even more complex programming environment. Agent-oriented architecture provides a suitable software environment for the development of an air combat simulation model based on the concept of rational agents. This approach allows the analyst to work at a high level, formulating concepts and aims, while keeping the detailed computer programming hidden.

1 Introduction Computer-based air combat modelling is a powerful tool, widely accepted for its usefulness. The extremely high cost of operating aircraft and their weapons has led to a rapid growth in the development and use of computer simulation models as a basis for tactics development, pilot training, and operational evaluation of weapon systems. However, models having the required level of delity are very costly to develop, not easily modi ed, and lack the ability to show the tactics being used. The analyst has to be able to program in the particular programming language used, and the tactics routines are not readily understood by pilots and operational planning sta. Present simulators combine a high degree of delity with the ability to simulate multiaircraft combat. These typically involve tens of thousands of lines of code and are dicult to manage and modify; combat scenarios involving multiple aircraft are especially dicult to model. The continuing rapid increase in speed of computing and advances in visual system hardware are allowing further signi cant advances to be made in simulation capability. However, these advances are dependent on further development and expansion of what is already a large body of software. The challenge is to provide improved performance whilst avoiding the existing shortcomings associated with large simulation codes. These are:  lack of exibility (in particular, team tactics are dicult to represent);  the time and cost associated with maintaining and modifying the simulation code;  the diculty of making modi cations without introducing errors;  the need to employ analysts who are also experienced programmers, and are familiar with the model and the code;  the air combat tactics being embedded in the model; and  the diculty in visualizing the tactics underlying the model as there is no intuitive means of describing them to pilots and operational personnel. The aims of the work reported here are to assist the analyst in making use of simulators for developing and evaluating new tactics, and to assist pilots and military planners to evaluate tactics prior to their operational introduction. These aims must be achieved in a way that makes use of the existing body of simulation software. They can be achieved by embedding the proposed new tactics representation in the current simulation environment, so that modi cations required in the code representing the physics of the model are minimal. A simulation model has a number of functional components:  a model of the aircraft and weapons physics;  a model of the aircraft sensors;  a model of the electronic warfare environment;  the air combat tactics;  user interfaces for both pilot and simulation control; and  a visual system to display the simulation. 1

The approach taken here relates to the representation, display and execution of air combat tactics by the simulation. The key is to separate the knowledge (tactics) from the physical model and visualization. This allows modi cation of the knowledge base without being concerned about the remaining code. Hence, the analyst need only work at the high level, formulating concepts and aims. This approach is eminently suited to dealing with extensive repertoires of pre-determined team tactics. The technology used is known as real-time procedural reasoning1. The current move to use of man-in-the-loop training simulations is accelerating the growth of computer simulation. Multi-aircraft full mission simulators will be perhaps the most demanding military application. High cost will restrict the number of pilot stations available, so that there will be a need to provide computer-generated surrogate pilots for both friendly and enemy aircraft.

2 Situation Awareness and Tactics Representation The tasks performed by a combat pilot can be broadly divided into two areas { situation awareness and tactics selection. Determination of the current situation is called situation awareness and the selection of appropriate actions in response to the situation is called tactics selection. Both stages require sophisticated reasoning and are closely linked. Having determined his current situation, a pilot needs to select and execute certain actions; these actions, in part, determine the next situation. A pilot's reasoning process can be characterized as consisting of beliefs about the real world, goals that need to be satis ed, mental plans or procedures that satisfy certain goals, and committed partial plans or intentions that a pilot has adopted in response to external events or internal goals. In situation awareness, a pilot must infer the beliefs, goals and intentions of other aircraft from their behaviour. In tactics selection, a pilot must react to his beliefs about the current situation or advance his goals towards a desired situation. While the problems of situation awareness and tactics selection are dicult for a single pilot in combat with a single enemy, the problems become far more complex when a team (of pilots) is in combat with an enemy team. The team as a whole needs to assess the situation by inferring not only the individual beliefs, goals, and intentions of enemy aircraft, but also the mutual beliefs, joint goals, and joint intentions of the entire enemy team. Similarly, tactics selection by a team is more dicult than the selection of tactics by a single pilot, because of the co-ordination and synchronization required.

3 Agent-Oriented Architecture A rational agent can be viewed as a system situated in the real world, continuously receiving perceptual input and responding by taking actions that aect the world. It can be characterized as having beliefs about the real world, goals that need to be satis ed, compiled plans or recipes for achieving certain goals, and committed partial plans or intentions that have been adopted in response to external events or internal goals. Compiled plans play an important role in such characterizations by side-stepping the classical planning problem. More recently, the logical and practical design of such rational 1 The term real-time is used here in the context of human behaviour and speed of decision making in a dynamic environment such as air combat, i.e., between one second and one minute.

2

agents have received increasing attention [Bratman et al., 1988; Cohen and Levesque, 1990; Rao and George, 1991]. The Procedural Reasoning System (PRS) [George and Lansky, 1986; Ingrand et al., 1992] is a situated system based on the BDI architecture that explicitly reasons about the beliefs, plans, goals, and intentions of a rational agent. The functional modules of the Procedural Reasoning System are given in Figure 1. In PRS, each agent is made up of the following components: Beliefs: The beliefs of an agent provide information about the state of the environment and are represented as ground rst-order logic formulas. Goals: The goals of an agent are descriptions of desired tasks or behaviours. The goal to achieve a certain formula P is written as (! P); to test for the truth of a formula is written as (? P); and to wait until a formula P is true is written (^ P). Plan Library: The plan library of an agent consists of a set of plans. Each plan is an abstract speci cation of possible sequences (and more complex orderings) of actions that can be used to accomplish certain goals or react to certain situations. Each plan consists of a name; an invocation condition, which speci es when a plan can be invoked; a precondition, which speci es the required belief state of an agent; and a body, which describes the possible sequences of actions as a rooted, directed, acyclic graph in which each edge is labelled with a primitive action or a goal expression of the above form. The plan library can also include metalevel plans, that is, information about the manipulation of the beliefs, goals, and intentions of the agent itself. These plans encode various methods for choosing among multiple applicable plans, modifying and manipulating intentions, and computing the amount of reasoning that can be undertaken, given the real-time constraints of the problem domain. Intention Structure: The intention structure contains a partially ordered set of tasks that the agent is committed to executing. These adopted tasks are called intentions. A single intention consists of some initial plan, together with all the sub-plans that are being used in attempting to successfully execute that plan. Intentions may be active or waiting for certain conditions to hold prior to activation. Some of these intentions may be metalevel intentions for decisions on various alternative courses of action. Interpreter: The interpreter manipulates the above components, selecting appropriate plans based on the system's beliefs and goals, placing them on the intention structure, and executing them. Unless some new belief or goal activates some new plan, the system will try to ful ll any intentions it has previously decided upon. This results in focused, goal-directed reasoning in which plans are expanded in a manner analogous to the execution of subroutines in procedural programming systems. Changes in the environment may lead to changes in the system's goals or beliefs, which in turn may result in the consideration of new plans that are not means to any already intended end. The system is therefore able to change its focus completely and pursue new goals when the situation warrants it. The system interacts with its environment, including other systems, through its database (which acquires new beliefs and discards old ones in response to changes in the environment) and through the actions that it performs as it carries out its intentions.

3

-

DATA INPUT

MONITOR

??

DATABASE (Beliefs)

QQ Q

6

KA LIBRARY (Plans)

 

SENSORS

INTERPRETER (Reasoner)

SYSTEM INTERFACES

6

 

GOALS DATA OUTPUT

QQ Q

ENVIRONMENT EFFECTORS

INTENTION STRUCTURE



 6

6

COMMAND - GENERATOR

Figure 1: Structure of the Procedural Reasoning System

4 Team-Oriented Architecture Just as a single agent can be considered a rational agent, a team is an organized collection of rational agents. A team can contain other teams as well. This results in a powerful organizational structure capable of modelling hierarchical as well as non-hierarchical social structures. In addition to individual beliefs, goals, plans, and intentions, a team of rational agents will also have: (a) mutual beliefs about the world and about each others' actions; (b) joint goals that need to be achieved collectively; (c) compiled joint plans (common to all team members) that specify the means of satisfying joint goals; and (d) committed joint plans or joint intentions, adopted as a response to a joint goal or an external event. Commitment to a joint plan may entail commitment to communicate to other members of the team. The system for air combat modelling consists of two types of agents { team agents and aircraft agents. One aircraft agent is created for each aircraft in the scenario. A team agent is created for each group of aircraft that forms a team. Figure 2 illustrates a 4v2 scenario where there are four \blue" aircraft and two \red" aircraft. The four blue aircraft are organized as two subteams, each with two aircraft. The two red aircraft also act as a team. The roles of leader and wingman are assigned to individual aircraft as well as to teams at the pre-mission-brief stage. The team structure can be understood by looking at the team agent BLUE1234, which consists of two team agents BLUE12 and BLUE34 with the former playing the role of the lead team and the latter playing the role of the wing team. The sub-team BLUE12 has two aircraft agents BLUE1, playing the role of the leader, and BLUE2, playing the role of the wingman. The creation of all these agents and their interactions with the simulator are handled by the agent COORDINATOR. Team agents and aircraft agents can be dynamically created, destroyed, and modi ed by the COORDINATOR. For example, if BLUE2 and BLUE3 are destroyed 4

W L RED12 R2 R1 L

L

W

W BLUE1234

B3

B1

B4 BLUE34 W

B2 BLUE12 L

Figure 2: Organization of Teams for a 4v2 Scenario in combat, the COORDINATOR can recon gure the agent structure so that the teams BLUE12, BLUE34, and BLUE1234 are replaced by the team agent BLUE14. Similarly, the roles of aircraft agents and team agents can be dynamically reassigned. Beliefs of each aircraft agent include: (a) heading; (b) position; (c) bearing to other aircraft; (d) range to other aircraft; and (e) headings of other aircraft. The positional information held by aircraft agents are egocentric. The beliefs of aircraft agents may vary depending on their role and their situation awareness. For example, an aircraft agent with the role of wingman might have beliefs about the position and heading of the lead aircraft agent. Also, the COORDINATOR can selectively pass range and bearing information with respect to other aircraft, thereby simulating the situation awareness of the agents. For example, the COORDINATOR might pass information regarding the range and heading of RED1 only to BLUE1 and not the others. This models situations where only BLUE1 is aware of RED1. A team agent also has beliefs of its own. These beliefs are about the team's heading, position, bearing, etc. Aircraft agents communicate their position and heading information to all their parents. For example, BLUE1's position and heading will be known to the team BLUE12 and the team BLUE1234. Thus information or beliefs ow up the hierarchical organization of agents. Aircraft agents tend to have low-level goals. These goals may have been adopted as a result of orders from higher level team agents or as subgoals to solve higher level goals. As a result, the plans of an aircraft agent tend to be low-level plans. The low-level plans include monitoring the current heading, changing the heading to acquire the desired heading, obtaining a certain lateral separation, ying at a certain heading for a given duration or distance, ying keeping a constant bearing to a given target, ring a missile, guiding a missile, etc. Note that none of the tactics appear as plans for the aircraft agents. The goals of team agents tend to be higher-level mission goals, like sites to defend or targets to attack. These goals are derived from the pre- ight mission brie ngs. Hence, team agents tend to have high-level tactical plans. These plans include the team searching for enemy aircraft in a particular formation, the team doing a pincer or cuto intercept, the team deciding that a particular sub-team is under threat and doing an evasion, the team choosing a target, etc. The top-level team adopts goals to perform intercepts, perform evasions, and search for enemy aircraft. Depending on the current scenario the team then chooses an appropriate plan to satisfy one of these goals which then becomes the intention of the team. The successful completion of this intention depends on the lower-level sub-teams adopting certain lower-level intentions. This is achieved by the team sending messages to the lower-level sub-teams to 5

(a)

(b)

(c)

Name: Achieve Pincer Intercept Invocation:

Name: Achieve Cuto Manoeuvre Team Invocation:

Name: Achieve Cuto Manoeuvre AC Invocation:

Precondition:

Precondition:

Precondition:

(*GOAL (! (pincer-intercept $enemy-team)))

(*GOAL (! (cuto-man $enemy-team $side)))

(*GOAL (! (cuto-man $enemy-team $side)))

(and (*FACT (this-agent $me)) (and (*FACT (this-agent $team)) (and (*FACT (this-agent $us)) (*FACT (single-aircraft $me))) (*FACT (:(single-aircraft $team))) (*FACT (team-left-right $us $left-team (*FACT (:(in-formation $lead-ac)))) Body: $right-team)))

Body:

k

1

Body:

(! (:(in-formation $us)))

? k

k

1 (! (do $lead-ac (! (cuto-man $enemy-team $side))

?k

2

2

(! (do-parallel ($left-team (! (cuto-man $enemy-team left)) ($right-team (! (cuto-man $enemy-team right)))) 3

? k

k

1 (! (control-until (have-separation $enemy-team $side) (in-roll-range $enemy-team)))

?k

2 (! (control-until (on-collision-course $enemy-team) (in-sort-range $enemy-team)))

?k

3

(! (attack-man $enemy-team))

? k

4

Figure 3: Joint Plans for Pincer Intercept and Cuto Manouevres

adopt certain intentions. The top-level team intention then gets suspended until it receives messages from the lower-level sub-teams that they have accomplished their tasks. In addition to the goals and intentions adopted as a response to higher-level team intentions, the individual aircraft can also adopt goals and intentions of their own to monitor their position, maintain bearing, re missiles, etc.

5 Sample Scenarios

5.1 Pincer and Cuto Intercepts

A pincer intercept by a team involves the team subdividing into two sub-teams, with each sub-team doing a cuto manoeuvre followed by an attack manoeuvre, in parallel. The toplevel plan to perform a pincer intercept is given by the plan Achieve Pincer Intercept (see Figure 3 (a)). A cuto intercept by a team involves the entire team doing a cuto manoeuvre to the left or right, followed by an attack manoeuvre. A cuto manoeuvre involves three phases, namely, obtaining a certain lateral separation from the enemy team's heading, ying straight till roll range, and then ying on collision course till sort range. An attack manoeuvre for a team involves each aircraft doing an attack manoeuvre, which involves ying in collision course and then ring and guiding the missile. The top-level plan to perform a cuto intercept is given by the plan Achieve cutoff intercept. If the target aspect is positive, the team does a cuto manoeuvre to the right, else it does a cuto manoeuvre to the left. There are two plans for a cuto manoeuvre, one for a team and the other for an aircraft. The plan Achieve Cutoff Manoeuvre Team (Figure 3 (b)) involves the lead aircraft of the team doing a cuto manoeuvre. The plan Achieve Cutoff Manoeuvre AC (Figure 3 (c)) involves the aircraft obtaining separation till roll range, followed by ying on collision course till sort range, when individual targets are chosen by each aircraft. There are three plans for an attack manoeuvre, one for a team to attack another team, the second for a team to attack a single aircraft, and the third for a single aircraft to attack 6

another single aircraft. Repeated execution of the rst and second plans will result in the team and the enemy team being sub-divided until a single aircraft has to attack a single enemy aircraft. This results in the third plan being executed. This plan involves the aircraft

ying on collision course till ring range, followed by ring the missile, followed by guiding the missile, followed by a y-through course till the missile has hit or missed. Figure 4 (a) shows a pincer intercept performed by a team of four blue agents against an opposing team of two red agents. The team agent BLUE12 is performing a cuto manoeuvre to the left and the team agent BLUE34 is performing a cuto manoeuvre to the right. R1

R2

(b)

(a)

R1

sort range

....................................................................

roll range

......................................................................

R2

maintain separation

.................................................................................

obtain separation

B1

B2

B3

B4

................................................................................. B1

B2

B3

B4

Figure 4: Pincer Intercept in a 4v2 Scenario

5.2 Sample 4v2 Scenario

In this section we consider a 4v2 scenario where there are four blue aircraft and two red aircraft. The hierarchical organization and roles of agents are as described earlier in Section 4 and shown in Figure 2. Consider the scenario shown in Figure 4 (a), where the team RED12 is ying straight onto the team BLUE1234. As the size of BLUE1234 is greater than the size of RED12, the team agent BLUE1234 decides to execute a pincer intercept with team agents BLUE12 and BLUE34 executing cuto manoeuvres. Aircraft agents BLUE1 and BLUE2 are obtaining separation towards the left of RED12's heading and BLUE3 and BLUE4 are obtaining separation towards the right of RED12's heading. The scenario after a few simulation cycles is depicted in Figure 4 (b). Now the team RED12 alters its course and decides to do a pincer intercept. The team agent BLUE1234 notices this 7

and instructs teams BLUE12 and BLUE34 to do a pincer intercept. This results in the BLUE12 team performing a pincer intercept and BLUE34 doing likewise. Thus a change in situation resulted in the team BLUE1234 changing its tactics. This change in tactics was possible because the interpreter of the team agent BLUE1234 is evaluating the situation at each and every cycle; as soon as there is any signi cant change in the situation it can modify its tactics and inform the sub-teams and aircraft agents to adopt alternative tactics. Similarly, as the interpreters in the sub-team agents and aircraft agents are continuously checking if they have received any orders from their higher level teams, they can adopt these orders as soon as they arrive. Also, the tactics, written as plans, are not written for each and every speci c scenario. The plans operate at the level of abstraction of a team and teams can consist of an arbitrary number of other teams or aircraft. The plans operate by binding the teams at run-time (based on the current scenario) and then sub-dividing the teams into sub-teams to carry out particular manoeuvres. As a result the same plans as mentioned above will work for any other scenario, like, 2v2, 4v4, 8v4, 6v8, etc.

6 Conclusion This paper presents a multi-agent real-time system that can represent and reason about joint mental attitudes such as mutual beliefs, joint goals, joint plans, and joint intentions. The approach outlined here makes it easier (and thus faster) to develop and modify tactics in air combat modelling. The programming complexity of multi-aircraft man-in-the-loop simulators demands such a conceptual breakthrough. Real-time procedural reasoning:  allows tactics to be constructed and displayed graphically (the analyst does not need to program the tactics in source code);  separates the tactics from the major body of the simulation code;  makes it easier to build, modify and display tactics, including team tactics;  makes simulated situations more easily understood by display of the underlying tactics involved; and  can be embedded within the simulator, thus taking advantage of existing simulator code. The system has a number of salient features which distinguishes it from other multiagent and distributed systems. Some of these are (a) it can represent and reason about abstract speci cations of team behaviour, namely, joint plans; (b) the actions performed by an agent that is part of a team are determined not only by its own individual mental state, but also the joint mental state of the team; (c) the agents can be grouped together as teams using a number of dierent organizational structures with speci c roles assigned to them; (d) the agents can reorganize themselves into dierent teams or change their roles based on changing situations; (e) the actions taken by the agents and the teams are based on their current situation and changing situations can lead to changes in the intentions of agents and teams. The above features make the system a powerful tool for reasoning about team behaviour and collaboratively solving problems, as well as providing a novel means for controlling air combat simulation models. A prototype of the team tactics representation and execution system, implemented using the Procedural Reasoning System, is running on Sun SPARCstations. PRS is written 8

in Common LISP and an extension of this system, called Multi-Agent Real-time System (MARS), is currently being written in C++. It will enable direct integration with conventional simulation software. The prototype team tactics system will then be ported to MARS, making it accessible to programmers who are not familiar with LISP.

Acknowledgments

The authors acknowledge the contributions of Dr. K. Anderson, Ms. S. Steuart and Mr. R. Muscat during the development of this system.

References [Bratman et al., 1988] M. E. Bratman, D. Israel, and M. E. Pollack. Plans and resourcebounded practical reasoning. Computational Intelligence, 4:349{355, 1988. [Cohen and Levesque, 1990] P. R. Cohen and H. J. Levesque. Intention is choice with commitment. Arti cial Intelligence, 42(3), 1990. [George and Lansky, 1986] M. P. George and A. L. Lansky. Procedural knowledge. In Proceedings of the IEEE Special Issue on Knowledge Representation, volume 74, pages 1383{1398, 1986. [Ingrand et al., 1992] F. F. Ingrand, M. P. George, and A. S. Rao. An architecture for real-time reasoning and system control. IEEE Expert, 7(6), 1992. [Rao and George, 1991] A. S. Rao and M. P. George. Modeling rational agents within a BDI-architecture. In J. Allen, R. Fikes, and E. Sandewall, editors, Proceedings of the Second International Conference on Principles of Knowledge Representation and Reasoning. Morgan Kaufmann Publishers, San Mateo, CA, 1991.

9