Intelligent Control for Obstacle Avoidance in the Self-Driving Car

• Published in: 2022 International Conference on Data Science and Intelligent Computing (ICDSIC) pp. 141 - 146
• Main Authors: Abd-Alrasool, Fatima M., Al-Moadhen, Ahmed A., Kamil, Haider G.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-11-2022
• Subjects: A algorithm; kinematics; model predictive control; Neural networks; obstacle avoidance; Path planning; Planning; potential field; Prediction algorithms; Real-time systems; Systems operation; Target tracking
• DOI: 10.1109/ICDSIC56987.2022.10076247

The keys to autonomous car accident avoidance in crucial traffic settings are motion planning and tracker control. It necessitates not just system operation but also reliable real-time. We merged an autonomous vehicle's motion planning and tracking control system in this paper to provide path planning and tracking for avoiding obstacles. The motion planning is based on a combination of A* and potential function algorithms. The Tracking Control Module is designed based on model predictive control (MPC) using a vehicle kinematics model and neural network to set the steering angle value. The simulation shows that the scheme can generate smooth paths that can be chosen as references to the console. In addition, it is shown by the results that the car in this method can track the reference path even at sharp angles. The maximum vehicle stopping error at the target is less than 0.1 m when the vehicle's desired speed is 40 km/h, and the maximum vehicle stopping error at the target is about 0.3 m when the desired speed of the vehicle is 120 km/h.