Road Map Assisted Multi-Target Tracking Method for Intelligent Vehicle

Accurate and real-time target tracking is essential to ensure the safety of intelligent vehicles. However, tracking in complex traffic environment is still a challenging problem. Considering that the road map is usually known in the traffic scene, this paper proposes a road map assisted Gaussian Mixture Probability Hypothesis Density (GM-PHD) filter to improve the target tracking performance. By the method, the road area and non-road area are extracted and distinguished from the map firstly, and the targets in the non-road area are removed; Then the number of birth target appearing in different areas is calculated at current time, to update the intensity information of birth target in the road area and non-road area iteratively. Finally, the updated birth target number is used as the state estimation of sensor measurement information at next time. The simulation results show that, compared with the Gaussian mixture probability hypothesis density filter without road map assistance, the proposed method improves the target tracking accuracy and reduce the calculation costs.