Collision Avoidance Using Behavioral-based Al Techniques
T. Gomi, Paula Volpe · 2005
This paper describes attempts made which have successfully created a set of algorithms which use behavior-based artificial intelligence (AI) techniques to collectively yield desirable collision avoidance behaviors in a dynamic operational environment. The algorithms consist of sensor- driven behavior modules which are necessary for dealing with various aspects of dynamic collision avoidance in urban traffic-like situations. These modules are implemented according to the principles of Subsumption Architecture, which is one of the key theories of behavior-based AI. A reduced-scale model operational environment was set up and the developed control system tested. Results indicate that it is feasible to construct control systems which allow fast and highly flexible maneuvering of vehicles operating in tight proximities. Outlines of the algorithms developed, as well as the experiments conducted, are discussed.