Cognition-inspired behavioural feature identification and motion planning ways for human-like automated driving vehicles

Human-like automated driving strategies could have advantages in traffic safety and comfort. However, the primary features of human-like driving behaviors are not clear yet. To deal with this problem, inspired by drivers' cognition way, a simple method is proposed to identify the critical parameters for human-like driving behaviors in extensive scenarios. Then with these parameters as terminal constraints, an interpretable motion planning method is developed in which the longitudinal and lateral planning units are coupled by some state information. With the driving data in the published literature, it is validated that the critical parameters match drivers' behaviors well. With experiments on the simulation and real-car platforms, it is validated that the motion planning method can generate diverse human-like behaviors in multiple scenarios.