Learning Long-Horizon Predictions for Quadrotor Dynamics

This github repository contains the codebase accompanying the paper Learning Long-Horizon Predictions for Quadrotor Dynamics (PDF) by Pratyaksh Prabhav Rao, Alessandro Saviolo, Tommaso Castiglione Ferrari, and Giuseppe Loianno.

An overview of our proposed methodology can be found video.

Abstract

Accurate modeling of system dynamics is crucial for achieving high-performance planning and control of robotic systems. Although existing data-driven approaches represent a promising approach for modeling dynamics, their accuracy is limited to a short prediction horizon, overlooking the impact of compounding prediction errors over longer prediction horizons. Strategies to mitigate these cumulative errors remain underexplored. To bridge this gap, in this paper, we study the key design choices for efficiently learning long-horizon prediction dynamics for quadrotors. Specifically, we analyze the impact of multiple architectures, historical data, and multi-step loss formulation. We show that sequential modeling techniques showcase their advantage in minimizing compounding errors compared to other types of solutions. Furthermore, we propose a novel decoupled dynamics learning approach, which further simplifies the learning process while also enhancing the approach modularity. Extensive experiments and ablation studies on real-world quadrotor data demonstrate the versatility and precision of the proposed approach. Our outcomes offer several insights and methodologies for enhancing long-term predictive accuracy of learned quadrotor dynamics for planning and control.

Citation

If you publish a paper with our codebase, please cite our paper published in IEEE/RSJ International Conference on Intelligent Robots and Systems:

@ARTICLE{longhorizondynamics2024,
  author={Rao, Pratyaksh and Saviolo, Alessandro and Ferrari, Castiglione, Tommaso and Loianno, Giuseppe},
  title={Learning Long-Horizon Predictions for Quadrotor Dynamics}, 
  booktitle={2024 IEEE/RSJ International Conference on Intelligent Robots and Systems},
  pages={},
  year={2024},
  organization={IEEE}

Installation

The code is tested with Python 3.8, PyTorch 2.12, and CUDA 11.8.

To install the dependencies, you can create a virtual environment with

conda create -n dynamics_learning python=3.8
conda activate dynamics_learning

Note: Install Pytorch and Pytorch Lighting based on your own system conditions. Here we use Linux and CUDA version 11.8.

Dependencies

git clone https://github.com/arplaboratory/FW-DYNAMICS_LEARNING.git
cd DYNAMICS_LEARNING
pip install -r requirements.txt

Data Preparation

mkdir resources
cd resources
wget https://drive.google.com/file/d/1BB-r63qgiqB5uJ5xVbTcfR-6j9rXCFCA/view?usp=sharing
unzip data.zip
rm data.zip

We first pre-process the dataset and store it in an hdf5 file -

cd ../scripts
python hdf5.py --dataset pi_tcn --history_length 20 --unroll_length 10

Note: Modify the dataset flag to specify pi_tcn dataset or neurobem dataset.

Training

Note: Modify the predictor_type flag to specify whether to train the velocity predictor or the attitude predictor.

python train.py --batch_size 1024 --model_type tcn --history_length 20 --unroll_length 10 --predictor_type velocity

Evaluation

python eval.py

Note: The evaluation script uses the best-trained model to assess performance on various test trajectories, reporting the velocity error for the velocity predictor and the quaternion error for the attitude predictor.

License

Please be aware that this code was originally implemented for research purposes and may be subject to changes and any fitness for a particular purpose is disclaimed. To inquire about commercial licenses, please contact Pratyaksh Prabhav Rao ([email protected]), Alessandro Saviolo ([email protected]), and Prof. Giuseppe Loianno ([email protected]).

    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.

    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.

    You should have received a copy of the GNU General Public License
    along with this program.  If not, see <https://www.gnu.org/licenses/>.