Implemention of generative AI with sampling optimizer for path planning for a mobile robot

• reimplemented with RVIZ visualiztion in closed loop with a rosbag and closed loop in pedsim gazebo simulation
• implemented generative AI for path planning for mobile robot in a crowded environment using VQ-VAE and PixelCNN

flowchart made by Naman Kumar

flowchart made by Soumo Roy

Built With

• Pedsim Gazebo Simulation
• ROS 1 noetic
• CUDA 11.8
• Pytorch

RVIZ window

• blue colour line is VQ-VAE + PixelCNN generated trajectory
• green colour line is PRIEST optimized trajectory
• green colour marker is the dynamic obstacles
• Simulation Demo

Pedsim Gazebo simulator

• Validated results in a crowded environment simulation to avoid humans (dynamic obstacles)
• Simulation Demo

How to run ?

  # this command is for running open loop using a rosbag
  conda env create -f environment.yml
  ./run_open_loop_rosbag.sh

  # this command is for running closed loop in pedsim gazebo simulation
  ./run_closed_loop_simulation.sh

Author

• Soumo Roy - soumoroy09@gmail.com
• Aadith Warrier (for guidance)

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Citation

Bibtex -

@misc{kumar2025crowdsurfersamplingoptimizationaugmented,
      title={CrowdSurfer: Sampling Optimization Augmented with Vector-Quantized Variational AutoEncoder for Dense Crowd Navigation}, 
      author={Naman Kumar and Antareep Singha and Laksh Nanwani and Dhruv Potdar and Tarun R and Fatemeh Rastgar and Simon Idoko and Arun Kumar Singh and K. Madhava Krishna},
      year={2025},
      eprint={2409.16011},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2409.16011}, 
}

Remark

This work is a re-implemtation of Crowdsurfer paper which was published at IEEE ICRA 2025 (A* star conference for robotics) at IIIT hyderabad under the guidance of Naman Kumar and Dr. Madhava Krishna