Simulating Elephant Behavior in a Virtual Environment

Abstract

The design of artificial systems or simulations inspired by biological behavior has recently attracted considerable interest. The field of animal behavior is diverse and may be studied from variety of perspectives. In the study of animal behavior in Zoology, in addition to the real experiments and field studies simulation experiments are also a useful source of knowledge and verification. There have been numerous incidents reported all over the world regarding attacking of agitated elephants. Giving a proper awareness of such incidents can be save valuable lives and resources. Therefore a simulation of elephant behavior is more important to improve the safety measures through the awareness. Motivated by this, the research presented in this thesis proposes a prototype for simulating the elephant behavior. Moreover, this model can be applied to many real world scenarios like military simulations, wild park designing, safety modeling, emergency preparation, evacuation modeling, as well as computer games and movies. There are two major components in the designed prototype in order to control behavior and actions of agents, Behavior Controller and Action Controller. Behavior Controller recommends the behaviors according to the environment, needs of agents and their cognition. Then Action Controller controls the actions in relation to the recommended behaviors. Since fuzzy logic is recognized as a powerful means to represent approximation intrinsic in human and animal reasoning and reacting, various components in the designed prototype are used fuzzy modeling. Steering behaviors implemented on OpenSteer library are used in order to simulate the behavior of agents. Performance evaluation is done for evaluating the implemented model. Profiling is used to analyse the performance of the specific functions with both elapsed and application inclusive times. Performance evaluation of the model is conducted using the machines with di erent configurations. It can be concluded that the configuration of the machines are not a ected for the performance for less number of agents. Since fuzzy model is the main component of the prototype, the performance of the fuzzy model is evaluated using di erent situations such as increasing the number of obstacles and increasing the threat level. The results are quite attractive and according to the results, it can be concluded that fuzzy logic modelling is a proper approach to control the unpredictable behaviors.