Kalman filter and vision localization based potential field method for autonomous mobile robots

In this paper, artificial potential field is utilized for autonomous mobile robot (AMR) path planning in dynamic uncertain environment. First, in order to obtain the repulsive force from the obstacles, sonar data are used in the AMR system to acquire the distance between the AMR and obstacles. And an improved Kalman filter is proposed in this paper to eliminate the sonar signal's disturbance due to the environmental noise. Then, by taking advantage of vision localization based on polynomial fitting, the attractive force towards the goal is obtained. The effectiveness of the proposed method for the AMR path planning is verified by experiments performed on the mobile robot, CASIA-1.