Development of a lane keeping steering control by using camera vanishing point strategy

A new computer vision strategy is developed by using the Vanishing Point from camera frames for lateral control. A vehicle kinematic model is derived based on a tricycle in order to analyze the system states in the control strategy. The roots loci of the system are analyzed to understand the influence of each state in the closed-loop. On each frame, two system states are extracted from the lane lines in the perspective view by using a vanishing point-based technique with the inverse perspective mapping. A state feedback controller is developed, and computational simulations are carried out in a Blender environment using a small car model. The validation of the proposed Vanishing Point technique is carried out by several computational simulations, which also confirm the effects of each state in the final controlled system response.