EricCousineau-TRI / passivity_cbf_demo

Passivity-based manipulator control with guaranteed constraint satisfaction in Drake

Passivity-Based Manipulator Control with Guaranteed Singularity Avoidance

This repository contains code for demonstrating a new approach to passivity-based control (PBC) of robot manipulators, which can guarantee both passivity and constraint satisfaction. We focus in particular on designing control barrier functions to avoid kinematic singularities.

Run the online interactive version on Google Colab here

Standard Constrained PBC	Proposed Approach

Dependencies

To run this code locally, you will need

• python3
• Drake, compiled with python bindings
• numpy
• matplotlib
• tkinter

Usage

Clone this repository: git clone https://github.com/vincekurtz/passivity_cbf_demo/

Start the Drake visualizer:

cd drake
./bazel-bin/tools/drake_visualizer

Run the main script: ./simulate.py

Additional options (different control schemes, plotting, etc) can be set in the "Setup Parameters" section of simulate.py.