Path Planning and Tracking of an Autonomous Car With High Fidelity Vehicle Dynamics Model and Human Driver Trajectories

This paper considers the problem of optimal trajectory generation for the realization of autonomous driving involving a human driven vehicle (HDV). The path-planning algorithm allows for cooperation between autonomous and human-driven vehicles by integrating the trajectory prediction of the human driver model (HDM) into the framework. Mixed-integer quadratic programming (MIQP) approach is used for path planning along with a linear model predictive controller (MPC) for path tracking. Computational simulations show the effectiveness of the proposed method for a human driven vehicle negotiating a stationary obstacle along with an autonomous vehicle on a three-lane road. A nonlinear vehicle model inclusive of rolling motions is developed and is subsequently linearized to be included as a plant model. The controller generates steering wheel and drive torque inputs. The steering and drive motor actuator constraints are incorporated in the control law. In addition to this a high fidelity multi-body dynamics model of the vehicle is developed in CarSim which emulates real vehicle characteristics. A co-simulation study is performed and the simulation results demonstrate effectiveness of proposed method.