Deep Learning for Path Tracking of Autonomous Mobile Robots

Path tracking is always the challenges in autonomous mobile robot. Relative motion of the robot relative to the path and disturbance from the surrounding environment affect image quality reducing the control performance. In this paper, a custom convolution neural network (CNN) is proposed for an autonomous mobile robot to follow the desired trajectory. A sufficient amount of data has been collected for training and learning the robot states. The proposed system is implemented by a low cost embedded mobile robot using Raspberry Pi. Experimental results have shown the efficiency of the proposed approach in terms of performance, accuracy and processing time.