A VRU Collision Warning System with Kalman-Filter-Based Positioning Accuracy Improvement

Pedestrian safety is currently one of the most popular and significant issues in the field of traffic safety. The GPS positioning error is recognized as one of the toughest obstacles to improve the current pedestrian collision warning and avoidance system. In this paper, a reliable collision warning system aimed at vulnerable road users is proposed, to detect the risk of traffic collisions and send the necessary warnings ahead of collisions happening. What's more, we also manage to reduce the GPS positioning error in the case of location acquisition with the help of the Kalman filter, which further increases the accuracy of the predicted outputs and warning messages. Finally, we conduct a few simulation tests, the results of which turn out to sufficiently verify the reliability of our system.