Simulation and Test of Vehicle Global Routing Planning Algorithm Based on Autonomous Driving Simulator

Abstract

Considering the driving scenarios of real autonomous vehicles and the simulation testing of the global routing planning algorithm, this study simulated and tested the global routing planning algorithm on structured roads and unstructured roads respectively. Under the structured road, parse the high-precision map information for autonomous driving, develop the information query interface, then use the lane connection relationship, vehicle driving intention and lane center-line sampling function to complete the global routing planning. Under the unstructured road, An improved A* algorithm based on dynamic grids, named Dynamic Lattice A* Route Planning(DLRP-A*) has been proposed, which adopts the idea of local dynamic and global dynamic grid division, and completes the global routing planning by iterative loop. Finally, this research built a simulation test link based on the autonomous driving simulator LGSVL (LG Silicon Valley Lab), and designed the communication module to implement the structured road planning algorithm to receive and send simulation message packages. Test results show that structured road planning algorithm is feasible. And under the unstructured road, the DLRP-A* algorithm proposed in this paper reduces the number of grid divisions compared to the global grid division in multiple test experiments. In addition, under the complex unstructured road, the DLRP-A* algorithm keeping the planning time while ensuring the planning routing accuracy.