Updated Hybrid A* Search for AGV Path Planning with Curvature Constraints

Overview

DESCRIBE PROJECT, SOURCE, ETC

NOTE: initial version roughly implemented the approach from the 2008 paper without respect to curvature, while updates integrated it properly and made strides to update various aspects for efficiency

Examples

• ADD VISUALIZATIONS AND STUFF HERE
• LINK TO COMPILED LATEX PDF AS DOCS

Building & Testing (C++ extension)

Prerequisites

• Python 3.10+
• (Optional) A C++ compiler with C++17 support
  • Linux: g++ or clang++
  • Linux (Debian/Ubuntu): sudo apt-get install build-essential python3-dev
  • macOS: Xcode Command Line Tools
  • Windows: "Build Tools for Visual Studio" (MSVC)

Python-first environment setup

From the repo root (DolgovMotionPlanner):

1. Create a virtual environment using Python's venv:

python -m venv .env

2. Activate the virtual environment:

a. For Linux users:

source .env/bin/activate

b. For Windows users:

.env/Scripts/activate

3. Finish environment setup and run initial tests with pytest:

a. Python only:

python -m pip install -U pip
pip install -e .
pytest -v

b. Optional C++ backend:

python -m pip install -U pip
pip install -e ".[cpp]"
$env:DOLGOV_BUILD_CPP="1" # on Linux: export DOLGOV_BUILD_CPP="1"
pip install -e . -v
pytest -m cpp

Troubleshooting

• ImportError: No module named hybrid_core

  • The extension wasn’t built. Re-run python -m pip install -e ..

• Windows/MSVC flag issues

  • setup.py selects MSVC-friendly optimization flags (/O2). If your environment still errors, remove extra_compile_args temporarily and try again

• Forcing a rebuild

  • Delete build/ and any *.so/*.pyd artifacts (or reinstall in a fresh virtualenv), then rerun python -m pip install -e ..

Roadmap

• MENTION SUBPAR C++ INTEGRATION AND PLANS TO UPDATE
• MISC IDEAS FOR IMPROVEMENTS