Path Planning in Unstructured Urban Environments for Self-driving Cars

We present a path planner for unstructured urban environments (PPUE) for self-driving cars. PPUE receives initial and goal poses as input, as well as maps of the environment. It employs a hybrid A* algorithm with two heuristics for generating paths, which are smoothed using Conjugate Gradient optimization. Different from previous works, PPUE uses: (i) an obstacle distance grid-map, instead of an occupancy grid-map, for representing the environment; and (ii) an accurate but simple collision model of the car. We have examined PPUE's performance experimentally in simulated and real world scenarios. Our results show that PPUE computes smooth and safe paths, which follow the kinematic constraints of the vehicle, fast enough for suitable real world operation.