Contenuto scaricabileScarica il pdf
Autonomous Vehicle (AV) With Accident Avoidance
According to Fatality Analysis Report System Web-Based Encyclopedia ( http://www-fars.nhtsa.dot.gov), there was 39,189 motor vehicle traffic crashed in United States in the year of 2006. Obstacle avoidance is an important problem in Artificial Intelligence research that could help us reduce the number of accidents. In this thesis, I discussed a reactive control approach of obstacle avoidance algorithm called wander for a static mover problem. This algorithm was embedded into a mini-car. I used a combination of Fuzzy Control and supervised learning paradigm of Neural Network to refine the control of the mini-car. I argued that the most important factors in avoiding obstacles are the speed of the car, the degree of turning of the car, and the speed of calculation of the input and output realization. I also compared my proposed solution with those addressed in several research projects and papers. At the end, I included the future improvements that can be done to perfect my solution.