An Expert System for Automated Highway Driving

The principal objective of this work is to study the applicability of expert systems to the task of guiding an automobile on a limited-access highway. The vehicle is assumed to be equipped with sensors detecting the surrounding traffic, road signs, and road geometry, as well as control logic and actuators governing the throttle, steering angle, and brakes. The goal of the expert system is to issue commands to the controllers, given the traffic situation, traffic signals, road signs, and the strategy chosen by the driver. The system presented here consists of a rule base providing the required driving knowledge, a backward-chaining inference engine that performs the reasoning, a knowledge-base compiler that optimizes the reasoning process, and a highway-traffic simulator that simulates vehicles on a highway, either controlled by a preset strategy or by an instance of the expert system.