Development of an Autonomous Vehicle Controller for Simulation Environments

Abstract

An autonomous vehicle controller for simulation and games environments must coordinate the movement and behavior of multiple vehicles in their chosen path (itinerary) in the given environment. This work presents the development of a controller that guides the movement and behavior of vehicles inserted in a 3D world, so they can respond accordingly to accomplish the given movement orders. Our proposed solution to this problem is mainly based on steering behavior, which were improved with adjustments and predictions to be able to better manage static and dynamic obstacles in the simulated environment. It also uses group movement models using communication and planning between agents as well as solutions to obstacle deviation in real time in order to avoid collisions between vehicles while maintaining the behavior as realistic as possible. We also use accelerating data structures to enable the solution to run in real time even in very large terrains with a large number of obstacles.