This repository contains a C++ implementation of the Hybrid A Star algorithm with a Python wrapper. It is used as one of the motion planning models in pylot, an erdos project.

Reference Paper: Practical Search Techniques in Path Planning for Autonomous Driving

Runtime varies heavily with the 'difficulty' of the scenario and the
discretization of search space. For example:

straight (first gif): ~ 1 ms
straight w/ obstacle (second gif): ~ 20 ms
turn w/ obstacle (third gif): ~ 3 s

git clone https://github.com/erdos-project/hybrid_astar_planner.git
./build.sh

There is a Python wrapper and C++ API. The Python wrapper is located in HybridAStar/hybrid_astar_wrapper.py and the C++ API is under src/HybridAStarWrapper.cpp. The following command will simulate a simple scenario to run the HybridAStar planning algorithm.

python3 HybridAStar/hybrid_astar.py