Abstract
An algorithm is introduced for vehicle navigation on roads and highways which is based on a virtual road partitioning concept. It is assumed that by using binary sensors to recognize the presence of the vehicle in cells of a virtual road matrix, this information can be made available to a communication center located near the road. Therefore, vehicles on the road can receive surrounding data from the center at every instant. This received information is fed into a two-layer expert fuzzy system in order to make decisions regarding the next motion of the vehicle. The first layer determines the state of a danger around the vehicle. The second layer in turn, determines the new path direction and speed for vehicle to ensure escaping from dangerous situations. The effectiveness of the method is examined for different scenarios in which the vehicle is expected to collide if no correcting action is undertaken. The results of simulations demonstrate suitability of maneuvers undertaken by the system to eliminate the danger of collision for automated-vehicles. Results vindicate the applicability of the new approach for collision avoidance of intelligent vehicles.
Keywords: motion planning, fuzzy logic, matrix road partitioning, collision avoidance.