A Path Following System for Autonomous Robots with Minimal Computing Power

Many existing path following approaches begin with a proposed solution and are then implemented on specialised hardware, which makes comparing approaches difficult. Many of these approaches also require a mathematical model of the robot or the track, a small change in the configuration of the robot or the track renders the entire solution inaccurate until the model is updated. This paper presents a method for autonomous path following using visual servoing. It differs from more conventional approaches by being designed to control an existing nonholonomic robot platform, by being designed to function on minimal processing hardware, and by being independent of both the robot platform and the dynamics of the track. The approach presented in this paper can be applied, with minimal modifications, to any autonomous robot platform with a vision sensor. Basic information about the path is extracted from each image, namely the lateral offset and local gradient of the track, and is used in a proportional steering controller.