Research on Lane-Change Online-Planning for Automotive Vehicles Based on Model Predictive Control

Shaojie Li,Lingyun Xiao,Shuo Cheng,Chao Huang,Bing Zhang,Xiang Chen

DOI: https://doi.org/10.1109/wrc-sara.2019.8931940

2019-01-01

Abstract:Lane change control is one of the key issues in the field of autonomous driving. Existing works have developed path-tracking controllers based on the pre-designed or off-line trajectory, which cannot adapt to the changes during the lane-change process. This paper adopts the real-time trajectory generation module and control module constructing the framework of lane change decision when there are at least 2 lanes on the left of the original lane. If the traffic environment changes, the vehicle will automatically update the trajectory to the left lane, redesign a new trajectory back into the original lane, or design a new lane into the 2nd left lane based on the real-time judgement of stability and safety criteria. It enables the vehicle to be more adaptable to unfavorable changes such as simultaneous lane changing maneuver of other vehicles. The online-planning module and the framework of the judgement criteria are the innovative points. Besides, this paper analyzes the influence of the prediction and control domains of Model Predictive Control. 20 steps and 10 steps were finally chosen as the prediction domain and control domain at the sample rate of 20Hz due to higher robustness and smaller hysteresis. The control module is based on the 3-degrees non-linear kinematic model. The discretization and linearization produce a linear model for model predictive control method. The co-simulation of Carsim and Matlab is conducted to verify the performance of proposed controller. In comparison with off-line trajectory planning methods, this method is characterized by shorter response time and stronger robustness in emergent conditions.