Online DNN-Driven Nonlinear MPC for Stylistic Humanoid Robot Walking with Step Adjustment

Giulio Romualdi, Paolo Maria Viceconte, Lorenzo Moretti, Ines Sorrentino, Stefano Dafarra, Silvio Traversaro, and Daniele Pucci
Co-first authors: Paolo Maria Viceconte and Giulio Romualdi

📅 Accepted for publication at the 2024 IEEE-RAS International Conference on Humanoid Robots (Humanoids), Nancy, France 🤖

📚 Paper     🎥 Video     🖼️ Poster     🔧 Experiments     🌐 Website     📂 Dataset

---

Reproducing the Experiments

You can reproduce the experiments using Docker, Conda, or Pixi.

Docker

Run the experiments via Docker for an isolated and reproducible environment.

1. Pull the Docker image:

   docker pull ghcr.io/ami-iit/dnn-mpc-walking-docker:latest

2. Launch the container:

   xhost +
   docker run -it --rm \
              --device=/dev/dri:/dev/dri \
              --env="DISPLAY=$DISPLAY" \
              --net=host \
              ghcr.io/ami-iit/dnn-mpc-walking-docker:latest

3. Wait for Gazebo to start and launch the experiment.

⚠️ Known Issue: The Gazebo real-time factor is scaled by a factor of 10 due to the MUMPS linear solver in the IPOPT Docker image. Alternative solvers (e.g., MA97) are available but cannot be redistributed.

---

Conda

Follow these steps to set up the experiments using Conda:

1. Install the environment:

   conda env create -f environment.yml

2. Activate the environment:

   conda activate dnn-mpc-env

3. Compile the code:

   cd paper_romualdi_viceconte_2024_humanoids_dnn-mpc-walking
   mkdir build && cd build
   cmake ..
   make -j
   make install

4. Run the simulation:

   ./run_simulation.sh

⚠️ The Gazebo real-time factor is scaled by a factor of 10 due to the MUMPS linear solver.

---

Pixi

To run the experiments with Pixi:

1. Clone the repository:

   git clone https://github.com/ami-iit/paper_romualdi_viceconte_2024_humanoids_dnn-mpc-walking
   cd paper_romualdi_viceconte_2024_humanoids_dnn-mpc-walking

2. Run the simulation:

   pixi run -e default run_simulation

Using MA97 Solver (Optional): If you have access to the Coin-HSL license, you can use the MA97 solver to improve performance:

1. Obtain the Coin-HSL archive (coinhsl-2023.11.17.zip) and place it in the ./coinhsl_src folder.
2. Run:

   pixi run -e coinhsl run_simulation

---

Maintainers

👨‍💻 Paolo Maria Viceconte

👨‍💻 Giulio Romualdi