Local Path Planning of Mobile Robots in Dynamic Unknown Environment Based on Prediction of Collision

The velocity and its distribution of obstacles is observed by an adaptive unscented Kalman filter, based on which the position of the dangerous region (potential collision region) is predicted. And based on its distribution, the dangerous region is expanded to ensure the security. By finding the best "free road" and estimate the "danger", a data reduction technique is employed to compress all the information needed, with the help of which the efficiency is improved significantly. Benefit from those the proposed algorithm gets a better performance in dynamic environment. The results of simulations have proved that.