Abstract: The autonomous driving technology is the core for the next generation transportation system. However, the mixed traffic streams at intersections are still present challenges, such as interaction with other non-autonomous vehicles, which must be addressed before implementation. Therefore, controlling and optimizing the turning trajectory at intersections is an important research field in autonomous driving. Based on two typical pre-crash scenarios of autonomous vehicles (AV), this study used the lattice planner algorithm (LPA) to handle turning trajectory for AV at intersections. LPA has three objectives: maximizing Time-to-collision (TTC) at intersections to enhance safety; minimizing jerk for comfortable ride; and increasing crossing speed for higher efficiency. Turning vehicle trajectory data were extracted from the Shanghai Naturalistic Driving Study (SH-NDS) for training, and then LPA was tested using CARLA simulator. Results demonstrate that LPA has high TTC, small jerk, and high crossing speed. Moreover, LPA learns the policies of human driving successfully and performs better compared to SH-NDS’s human driving data. These indicators demonstrate that AV can use LPA to handle trajectory planning and controlling for crossing intersections.
Linjia He, Xuesong Wang*, Han Chen, Jie Wang. Application of Lattice Planner Algorithm to Handle Turning Trajectory for Autonomous Vehicles at Intersections. Transportation Research Board 101st Annual Meeting, Washington D.C., USA, 2022. 1.9-13.