The collision avoidance (CA) system is a pivotal part of the autonomous vehicle. Ability to navigate the vehicle in various hazardous scenarios demands reliable actuator interventions. In a complex CA scenario, the increased nonlinearity requires a dependable control strategy. For example, during collisions with a sudden appearing obstacle (i.e. crossing pedestrian, vehicle), the abrupt increment of vehicle longitudinal and lateral forces summation during the CA maneuver demands a system with the ability to handle coupled nonlinear dynamics. Failure to address the aforementioned issues will result in collisions and near-miss incidents. Thus, to solve these issues, a nonlinear model predictive control (NMPC)-based path tracking strategy is proposed as the automated motion guidance for the host vehicle CA architecture. The system is integrated with the artificial potential field (APF) as the motion planning strategy. In a hazardous scenario, APF measures the collision risks and formulates the desired yaw rate and deceleration metrics for the path replanning. APF ensures an optimal replanned trajectory by including the vehicle dynamics into its optimization formulation. NMPC then acts as the coupled path and speed tracking controller to enable vehicle navigation. To accommodate vehicle comfort during the avoidance, NMPC is constrained. Due to its complexity as a nonlinear controller, NMPC can be time-consuming. Therefore, a move blocking strategy is assimilated within the architecture to decrease the system’s computational burden. The modular nature of the architecture allows each strategy to be tuned and developed independently without affecting each others’ performance. The system’s tracking performance is analyzed by computational simulations with several CA scenarios (crossing pedestrian, parked bus, and sudden appearing moving vehicle at an intersection). NMPC tracking performance is compared to the nominal MPC and linear controllers. The effect of move blocking strategies on NMPC performance are analyzed, and the results are compared in terms of mean squared error values. The inclusion of nonlinear tracking controllers in the architecture is shown to provide reliable CA actions in various hazardous scenarios. The work is important for the development of a reliable controller strategy for multi-scenario CA of the fully autonomous vehicle.
Bio-mimetic trajectory generation of robots via artificial potential field with time base generator
