Simultaneous Realization of Planning and Control for Lane-change Behavior Using Nonlinear Model Predictive Control

The demand for smart control technology is glowing according to the increase of expectation for autonomous driving. This paper presents an integrated planning and control scheme for automated lane-change behavior based on real-time nonlinear model predictive control (NMPC). In the proposed method, switched weighting parameters are used in the cost function over the prediction horizon. This enables us to reduce the number of reference points for control compared with the conventional framework. In addition, some safety constraints, such as collision avoidance and/or speed limitation are considered in a consistent manner. Although the NMPC usually requires huge computational cost, by applying the Continuation/GMRES (C/GMRES) method, it can be drastically reduced and become possible to be implemented in real time. Finally, validity of the proposed method is demonstrated by simulation study.