SLIP-based Iterative Learning for Efficient and Compliant Locomotion of Articulated Soft Quadrupeds

This repository contains the implementation of a novel control framework for soft quadruped robots, combining SLIP (Spring Loaded Inverted Pendulum)-based trajectory optimization with iterative learning control to achieve efficient and compliant locomotion.

This work was developed as part of my MSc thesis at TU Delft and has been published in the IEEE Robotics and Automation Letters (RA-L).

📄 Read the Thesis | 📰 Published Paper (IEEE RA-L)

Overview

The framework consists of two main components:

1. Offline Trajectory Planner: Generates energy-efficient reference trajectories using SLIP-based models
2. Iterative Learning Controller: Learns to track the reference through experience, improving tracking performance over iterations

Installation

Create and activate the conda environment:

conda env create -f environment.yml
conda activate locomotion

Note: Newer versions of matplotlib may cause errors when plotting optimization results.

Usage

Running the Trajectory Planner

python planner.py

Running the Controller

python controller.py

Citation

If you use this code in your research, please cite:

@ARTICLE{10313054,
  author={Ding, Jiatao and Sels, Mees A. van Löben and Angelini, Franco and Kober, Jens and Santina, Cosimo Della},
  journal={IEEE Robotics and Automation Letters}, 
  title={Robust Jumping With an Articulated Soft Quadruped Via Trajectory Optimization and Iterative Learning}, 
  year={2024},
  volume={9},
  number={1},
  pages={255-262},,
  doi={10.1109/LRA.2023.3331288}}