Keisuke Uto

Appropriate overtaking motion generating method using predictive control with suitable car dynamics

We developed an appropriate overtaking motion generating method suitable for autonomous cars using model predictive control (MPC). The MPC framework determines the control input in accordance with optimization that reflects the prediction of the state transition for a finite period. Some related studies of motion planning using MPC presented simplified optimization methods that can be used in almost real time. However, substantially simplified optimization often causes differences in behavior from actual cars. Our system consists of a method for planning acceleration and the steering change ratio that can generate accurate overtaking motion. We developed this method to solve the problem of overtaking. The applicability of our system to autonomous cars was evaluated using a car dynamics simulator.