machines-in-motion/reactive_planners

Readme

Contains a list of reactive planners specialized in locomotion of legged robots. The reactive planner adapts the step location and timing of the gait based on feedbck from the CoM states and sends the desired swing foot trajectories to an instantanous controller for tracking.

Installation

Standard dependencies

Here all the pip and apt install-able stuff

Download the package

Install treep and colcon .

Build the package

Then follow the instructions below:

# install treep and colcon
pip install -U treep colcon-common-extensions
# change directory to your devel folder
mkdir devel
cd devel
# Clone the treep configuration
git clone https://github.com/machines-in-motion/treep_machines_in_motion.git
# Clone the code base
treep --clone REACTIVE_PLANNERS
# go and build the code
cd workspace
colcon build --cmake-args -DCMAKE_BUILD_TYPE=Release
# source the environment for using the code
source install/setup.bash

Usage

Demos/Examples

To run Bolt walking in simulation: python3 src/reactive_planners/demos/demo_reactive_planners_bolt_step_adjustment.py

To run Solo12 walking in simulation: python3 src/reactive_planners/demos/demo_dgh_sim_solo12_step_adjustment.py

Reference

This package contains the implementation of the algorithms depicted in:

Elham Daneshmand, Majid Khadiv , Felix Grimminger and Ludovic Righetti. “Variable Horizon MPC With Swing Foot Dynamicsfor Bipedal Walking Control.”, IEEE Robotics and Automation Letters, 6(2). https://arxiv.org/abs/2010.08198 (2021)
Majid Khadiv, Alexander Herzog, S. Ali A. Moosavian and Ludovic Righetti. “Walking Control Based on Step Timing Adaptation.”, IEEE Transactions on Robotics, 36(3). https://arxiv.org/abs/1704.01271 (2020)