ABSTRACT
ABSTRACT: This paper presents an extended formulation to the obstacle avoidance issue for autonomous ground vehicles presented in [1], integrating new functionalities such as multiple and moving obstacles avoidance. Planning and tracking tasks are split into two levels of a hierarchical controller. The high level generates a feasible and safe path solving a nonlinear model predictive control problem using a nonlinear vehicle model defined by spatial coordinates where vehicle position along a path is used as the independent variable. The low level tracks the planned path using a four-wheels vehicle model and solving an MPC problem with a sequential quadratic programming approach. Simulations show the capability of the extended high level control logic to properly detect feasible and obstacle free paths with multiple and moving obstacles on the road where their future trajectories are considered “a priori” known
Keywords: Safety path planner, collision risk estimation, autonomous vehicle, obstacle avoidance, Hierarchical MPC
1 INTRODUCTION
Today modern technologies have the potential for improving safety in the automobile industry: in fact modern vehicles are equipped with computational, sensing and actuating capabilities allowing the design of Advanced Drive Assistant Systems (ADAS), such as Adaptive Cruise Control (ACC) and lane-keeping systems [2]. Current trend in the development of ADAS is focused on systems able to work completely autonomously in complex environments.
