Autonomous Emergency Collision Avoidance and Stabilisation in Structured Environments (2004.07987v1)

Abstract: In this paper, a novel closed-loop control framework for autonomous obstacle avoidance on a curve road is presented. The proposed framework provides two main functionalities; (i) collision free trajectory planning using MPC and (ii) a torque vectoring controller for lateral/yaw stability designed using optimal control concepts. This paper analyzes trajectory planning algorithm using nominal MPC, offset-free MPC and robust MPC, along with separate implementation of torque-vectoring control. Simulation results confirm the strengths of this hierarchical control algorithm which are: (i) free from non-convex collision avoidance constraints, (ii) to guarantee the convexity while driving on a curve road (iii) to guarantee feasibility of the trajectory when the vehicle accelerate or decelerate while performing lateral maneuver, and (iv) robust against low friction surface. Moreover, to assess the performance of the control structure under emergency and dynamic environment, the framework is tested under low friction surface and different curvature value. The simulation results show that the proposed collision avoidance system can significantly improve the safety of the vehicle during emergency driving scenarios. In order to stipulate the effectiveness of the proposed collision avoidance system, a high-fidelity IPG carmaker and Simulink co-simulation environment is used to validate the results.