Vision based lane reference detection and tracking control of an automated guided vehicle

In this works, a novel lane reference calculation approach is developed in order to define a safe and optimal trajectory to be tracked by a road vehicle. In a first step of the design, the lane boundaries are extracted using a vision based lane detection algorithm, which consists in an homography transformation that helps to extract features from the captured image frames in order to map a set of control points needed to reconstruct the lane path edges. In a second step, a real-time safe optimal reference trajectory is generated that explicitly takes into account the kinematics of a four-wheel road vehicle, a feasible trajectory delimited by the estimated lane boundaries is therefore generated, the planned trajectory fed to the vehicle controller enabling the road vehicle to tracked it. In order to have fast tracking of the reference trajectory, a non-singular sliding mode controller is implemented to achieve finite convergence of the tracking errors. The whole algorithm is applied on a four wheels automated guided vehicle model. Numerical simulations are presented to validate the proposed approach.