BugWorld: A distributed environment for the development of controlarchitectures in multi-agent worlds

Abstract

This software allows the simulation of several autonomous agents (simulated robots or simply bugs) equipped with distance sensors. Each bug can define a unique set of distance sensors located in the center of its circular body. The simulator consists of several different processes. Each new agent is a separate process that communicates with the simulator via Internet. In this way a great many of agents can be simulated at the same time on different workstations. The process that maintains all the information about the simulated world can run on a separate machine and is not slowed down by agents that might need more and more CPU time as the complexity of their architectures increases. Contents 1 Motivation 2 2 Concept of the distributed simulator 2 3 Implementation 3 4 Genetic Algorithm connection 5 5 Usage of the simulator 5 6 Future enhancements 8 7 Conclusion 10 1 Motivation There is a growing doubt that the traditional artificial intelligence (AI) approach to intelligent behavi...