Collaborative Behavior Modeling And Simulation Of Groups

Collaborative behavior modeling and simulation (M&S) of military groups is notonly an important branch of the military M&S field, but also the theoretical foundation,one important method and a cutting edge of computer-generated forces. Large-scalemilitary group behavior modeling is extremely difficult due to the complex organizationand command relationships between groups, as well as cooperative behaviors andindependent behaviors of them sponsored senior intelligence.At present, the methods of the individual behavior modeling and simulation,represented by agent technology, have been extensively studied. However, thesemethods, which are mainly proposed for individual forces or certain specific issues inthe behavioral modeling, cannot be applied to the modeling and simulating of groupbehaviors directly, for the lack of description of high-level cooperative relationshipsbetween groups. Existing group collaborative behavior modeling methods can bedivided into two categories: bottom-up ways centering on behavior modeling ofindependent virtual entities; and top-down ways centering on behavior modeling oforganizational cooperative relationships. The former focus on the individual’sperception and interactions, without integrate organizational collaborative behavioralmodels, so it is difficult to depict the group behaviors accurately and furthermoreinextensibly. The latter however can overcome the deficiencies of the former, and nowhas become an promising direction in the group behavior modeling area.In order to meet the requirements of large-scale military combat simulation, tacticsanalysis and evaluation, large-scale military operations other than war (MOOTW)simulation and the presuppose scenarios analysis and evaluation, this thesis studies thetheory of organizational structures, task planning and behavior generating concerningthe military group collaborative behavioral modeling problem, as well as the realizationand implement technologies. Finally, the theories and implementation technologies aretested and verified on typical cases of joint operations and MOOTW.This major work and innovations of the thesis are as follows:(1) A method named GBDM(Group Behavior Describing Method), combining thedomain ontology and the description logic of tasks together, is proposed for thecollaborative behavior modeling of complex military groups. This method combines theknowledge representation of the domain ontology with the task semantics of thedescription logic of tasks to provide precise descriptions and reasoning of concepts. It ischaracterized as accuracy semantics and outstanding expressing, and ensure that thecomputer representation of the group behavior simulation can reflect the user’s ideasabout deployments, actions and encounters of the operations reliably. Whereas,traditional simulation systems are object-oriented, and doesn’t have the abilities of semantic representation or reasoning. In the meantime, the GBDM method provideswell formal reasoning support for scenarios validation, concepts inference, and also thededucing of role norms and abilities.(2) A formal model of the task planning based on the GBDM method is given.With this model, a decision theory of the achievability of task planning and its deducingalgorithms are proposed.In this dissertation, a task planning method named TPDLT(Task Planning based onthe Description Logic of Tasks) based on GBDM is proposed for joint operations andMOOTW. TPDLT allows planning of multilayer groups to be described in aformalizing way. It forms a whole theoretical system and provides a formalizeddescription of the task planning in joint operations and MOOTW. Based on thedescription logic of tasks, we also gives a definition of the achievability of task plans, adecidable theorem of the achievability under joint strategies. and a deducing algorithmfor task planning besides. As illustrated by a case study, with the TPDLT method, thetask interaction semantics in virtual groups can be described succinctly and accurately,and the task planning in multi-scale and multi-resolution modeling can also be welldescribed. In the meantime, it provides powerful service for automatic deducing andvalidation of the achievability deduce of tasks. The efficiency of the development andvalidation for users’ scenarios can be greatly improved with this method.GBDM allows TPDLT to avoid the frame problem which most planning logicsmay encounter.(3) A hierarchical model named GSCP with complete formal semantics is proposedto meet the demands of computer generated group forces. The GCGF model consists ofan organizational structure model GRNA and a forces behavior model GSCP. GRNAdescribes the high-level structure of organizational groups, while GSCP describes thelow-level behaviors. GRNA divides the group structure into two parts: cognitivecomponent and interactive component. The cognitive component contains descriptionabout organizational groups and their capabilities, while the interactive componentcontains the roles played by the groups and the norms to be obeyed. GRNA can expresshigh-level interactions, and has a complete and reliable axiomatic system. GSCP, whichis proposed for the behavior and performance modeling of large scale groups forces,combines the ideas of cybernetics and swarm intelligence, and supports group behaviormodeling on three categories including hierarchical synchronization, mutualsynchronization and self-synchronization.(4) A behavior verification and validation method called ASM-VV is proposedbased on abstract state machine to meet the demonds of behavior modeling inlarge-scale joint operations and MOOTW, for behaviors in these fields are verycomplex and hard to design, verify and validate. In the method, the behavior of CGF isformally described by abstract state machine language, and it can be used for strictly verification and validation. Illustrated by some examples, the method can not only helpusers check the completeness of their designs before implemention, but also helpchecking the soundness of the designs and validating the correctness of theimplementation.(5) An engineering approach for collaborative behavior modeling of militarygroups is proposed, and a prototype system is developed based on it. In this prototype,an Organizational Group Behavior Markup Language (OGBML) is designed andrealized. OGBML is an XML-based language and can give a full description for thecollaborative behavior model of groups. This prototype also provides some flexibletools like scenario manager, model manager, simulation manager etc.. These tools helpdevelopers create their own cooperative behavior simulation system easily.Finally, the feasibility and the validity of our work are illustrated and verified bytwo applications of joint operations and MOOTW.