ALX: Autonomous Vehicle Guidance for Roadway Following and Obstacle Avoidance

摘要

The report presents results of the research performed on the Autonomous LandExperimental Vehicle (ALX) at the University of Minnesota. ALX autonomously follows roadways through the use of visual perception, and executes obstacle detection and collision avoidance through the use of ultrasonic sonar range sensors. The report describes the ALX embedded real-time control system based on a multi-processor, multi-taking architecture, and presents algorithms used for visual perception, path tracking, position estimation, obstacle detection, and collision avoidance. Computer simulation and experimental results also are presented. The main impetus for the development of AVCS is safety. The National Traffic Safety Board has indicated that in excess of 40% of heavy truck accidents are due to driver fatigue. Technologies that provide for collision avoidance and lane following can potentially 'reduce urban and rural freeway accidents by a minimum of 30 to 35 percent, respectively. Accident reductions of this magnitude would eliminate approximately 71,000 accidents per year and $700 million in accidents costs.