Spatio-temporal risk prediction-based longitudinal motion control for autonomous vehicles in blind zone environments

Blind zone environments cause collision risks for autonomous vehicles (AVs), and how to achieve safe and efficient blind zone passing is an important challenge. In this paper, a longitudinal motion control strategy based on the spatio-temporal risk prediction (STRP) and model predictive control (MPC) algorithm, named MPC-STRP, is investigated for AVs in the blind zone. Before detecting the obstacle in the blind zone, the concept of virtual obstacle is proposed, and its corresponding spatio-temporal risk field is established. Upon detection of the obstacle, the real obstacle spatio-temporal risk field is established to replace the virtual one. On the basis of the spatio-temporal risk field and the obstacle motion model, the STRP is developed to characterize the risks that exist in the blind zone in a future period. The longitudinal motion controller is comprised of an upper-level controller that employs the MPC-STRP to determine the optimal longitudinal acceleration, and a lower-level controller that calculates the torque at the front and rear wheels. The simulation results demonstrate that the longitudinal control strategy based on MPC-STRP proposed in this paper can effectively ensure the safety of the AV when passing through the blind zone while maintaining the passing efficiency and comfort.