The source code on this repository is dependent on the do-mpc library https://www.do-mpc.com/en/latest/

BibTex

@article{shamsah2024socially,
  title={Socially acceptable bipedal robot navigation via social zonotope network model predictive control},
  author={Shamsah, Abdulaziz and Agarwal, Krishanu and Katta, Nigam and Raju, Abirath and Kousik, Shreyas and Zhao, Ye},
  journal={IEEE Transactions on Automation Science and Engineering},
  year={2024}
}

@article{shamsah2024real,
  title={Real-time Model Predictive Control with Zonotope-Based Neural Networks for Bipedal Social Navigation},
  author={Shamsah, Abdulaziz and Agarwal, Krishanu and Kousik, Shreyas and Zhao, Ye},
  journal={IEEE/RSJ International Conference on Intelligent Robots and Systems},
  year={2024}
}

Structure

The examples folder contains a number of different examples with different models and constraints setup. Each example contains three main codes.

• main_offline.py

  • runs the MPC loop
  • sets initial conditions
  • saves the results

• template_mpc.py

  • sets cost function
  • time varying paramters (pedestrians positions, terrain, obstacles, goals), MPC horizion, and sets the constrains

• template_model.py

  • sets system model
  • sets system states
  • sets constraints equations

(For more details on the structure of the code visit https://www.do-mpc.com/en/latest/)

examples

• decoupled

  • SZN-MPC decoupled

• coupled

  • SZN-MPC coupled

Running the code

run python main_offline.py, it will save the results in the results folder, after it runs for the specified number of steps.

cd ~/decoupled
python main_offline.py

Visualization

cd ~/decoupled
python animate.py