FFRENETIX Motion Planner & Multi-agent Scenario Handler

Welcome to the TUM FRENETIX Motion Planner. Here you find the modules, the code and the information to run our high performance motion planning algorithm for autonomous driving tasks.

📖 Overview Modules

This repository includes a Frenet trajectory planning algorithm and a Multi-agent Simulation Framework in the CommonRoad scenario format. The trajectories are generated according to the sampling-based approach in [1-5] including two different implementations. The Repo provides a python-based and a C++-accelerated Motion Planner Frenetix implementation. The multi-agent simulation can be used to integrate and test different planning algorithms. FRENETIX is an modular and adaptive motion planning environment that allows researchers to add and exchange the following modules:

Detailed documentation of the functionality behind the single modules can be found below.

1. General Motion Planning Algorithm

2. Frenetix C++ Trajectory Handler

3. Commonroad Scenario Handler

4. Module M2: Behavior Planner

5. Module M3: Occlusion-aware Module

6. Module M4: Trajectory Prediction: Wale-Net

7. Module M5: Risk-Assessment

8. Module M6: Reinforcement Learning Module Extension

🔧 Requirements & Pre-installation Steps

Requirements

The software is developed and tested on recent versions of Linux. We strongly recommend to use Ubuntu 22.04 or higher. For the python installation, we suggest the usage of Virtual Environment with Python 3.11, Python 3.10 or Python 3.9 For the development IDE we suggest PyCharm

Pre-installation Steps

1. Make sure that the following dependencies are installed on your system for the C++ implementation:

   • Eigen3
     • On Ubuntu: sudo apt-get install libeigen3-dev
   • Boost
     • On Ubuntu: sudo apt-get install libboost-all-dev
   • OpenMP
     • On Ubuntu: sudo apt-get install libomp-dev
   • python3.11-full
     • On Ubuntu: sudo apt-get install python3.11-full and sudo apt-get install python3.11-dev

2. Clone this repository & create a new virtual environment python3.11 -m venv venv

3. Install the package:

   • Source & Install the package via pip: source venv/bin/activate & pip install . or poetry install
   • Frenetix should be installed automatically. If not please write [email protected].

4. Optional: Download additional Scenarios here

🚀🚀🚀 Frenetix-Motion-Planner Step-by-Step Manual

1. Do the Requirements & Pre-installation Steps

2. Change Configurations in configurations/ if needed.

3. Change Settings in main.py if needed. Note that not all configuration combinations may work. The following options are available:

   1. use_cpp: If True: The C++ Frenet Implementations will be used.
   2. Set the scenario name you want to use.

4. Run the planner with python3 main.py

5. Logs and Plots can be found in /logs/<scenario_name>

🚗🛣️🚙 Multi-agent Simulation Framework

Run Multi-agent Simulation

1. Do the Requirements & Pre-installation Steps
2. Change Configurations in configurations/ if needed.
   By default, a multi-agent simulation is started with all agents.
   The multi-agent simulation settings can be adjusted in configurations/simulation/simulation.
3. Change Settings in main_multiagent.py if needed
   1. Set the scenario name you want to use.
   2. evaluation_pipeline: If True: Start an evaluation pipeline with all scenarios
4. Run the simulation with python3 main_multiagent.py
5. Logs and Plots can be found in /logs/<scenario_name>

Integration of external Trajectory Planner

1. A base class with all attributes necessary for the simulation is provided in cr_scenario_handler/planner_interface
2. Create a new file with an interface to fit your planner and save it in cr_scenario_handler/planner_interface
   The new interface must be a subclass of PlannerInterface.
3. In configurations/simulation/simulation adjust used_planner_interface with the class-name of your interface

🚸 Occlusion-aware Module

Also checkout the external Occlusion-aware Module here.

🤖 Reinforcement Learning Framework

Also checkout the external Reinforcement Learning Agent Framework here.

📈 Test Data

Additional scenarios can be found here.

📇 Contact Info

Rainer Trauth, Institute of Automotive Technology, School of Engineering and Design, Technical University of Munich, 85748 Garching, Germany

Marc Kaufeld, Professorship Autonomous Vehicle Systems, School of Engineering and Design, Technical University of Munich, 85748 Garching, Germany

Johannes Betz, Professorship Autonomous Vehicle Systems, School of Engineering and Design, Technical University of Munich, 85748 Garching, Germany

📃 Citation

The whole FRENETIX setup is further explained in this video If you use this repository for any academic work, please cite our code:

• Analytical Planner Paper

@ARTICLE{Frenetix,
  author={Trauth, Rainer and Moller, Korbinian and Würsching, Gerald and Betz, Johannes},
  journal={IEEE Access}, 
  title={FRENETIX: A High-Performance and Modular Motion Planning Framework for Autonomous Driving}, 
  year={2024},
  volume={},
  number={},
  pages={1-1},
  doi={10.1109/ACCESS.2024.3436835}
  }

• Multi-agent Simulation Framework

@INPROCEEDINGS{multiagent2024,
  author={Kaufeld, Marc and Trauth, Rainer and Betz, Johannes},
  booktitle={2024 IEEE Intelligent Vehicles Symposium (IV)}, 
  title={Investigating Driving Interactions: A Robust Multi-Agent Simulation Framework for Autonomous Vehicles}, 
  year={2024},
  volume={},
  number={},
  pages={803-810},
  doi={10.1109/IV55156.2024.10588423}
}

• Occlusion-aware Planning

@ARTICLE{FRENETIX_Occlusion,
  author={Trauth, Rainer and Moller, Korbinian and Betz, Johannes},
  journal={IEEE Open Journal of Intelligent Transportation Systems},
  title={Toward Safer Autonomous Vehicles: Occlusion-Aware Trajectory Planning to Minimize Risky Behavior},
  year={2023},
  volume={4},
  number={},
  pages={929-942},
  doi={10.1109/OJITS.2023.3336464}
  }