react-control
This repository contains a controller for the iCub for performing reaching with simultaneous reactive whole-body obstacle avoidance. The module thus encapsulates the functionality of an inverse kinematics solver and the actual controller that generates the commands to reach a position in Cartesian space. The obstacles - perceived visually or through the iCub's sensitive skin - are incorporated on the run as additional constraints.
The reaching controller is standalone in the reactController module here, but to acquire the information about the visual/tactile obstacles, it relies on communication with modules from the peripersonal-space repository.
Structure of the repository
The core module is the reactController.
The old modules and tests which had an internal/development character are placed in deprecated folder.
Documentation of the reactController
Dependencies:
yarp
iKin
skinDynLib
ctrlLib
osqpEigen
Eigen3Principles of operation:
This modules parallels the functionality of the iCub Cartesian interface, but both the solver and controller are encapsulated in the single problem formulation.
The task is formulated as a quadratic program that seeks joint velocity commands respecting multiple constraints while minimizing the magnitude of the velocities. The obstacles are dynamically managed by mapping visual and proximity (pre-collision) and tactile (post-collision) obstacles as additional constraints to the motion controller. The solution to the task (given in the joint velocity space) is integrated into new joint positions sent directly to the robot (see the schema below). Importantly, there is no distinction between end-effector and whole-body: the obstacles affect every body part in the same way - by limiting the joint velocities.
Citation
If you found this work useful, please cite our manuscript:
@ARTICLE{rozlivek2024harmonious,
author={Jakub Rozlivek and Alessandro Roncone and Ugo Pattacini and Matej Hoffmann},
journal={IEEE Transactions on Robotics},
title={{HARMONIOUS – Human-like reactive motion control and multimodal perception for humanoid robots}},
year={2024},
volume={?},
number={?},
pages={???},
doi={10.1109/TRO.2024.3502216}