SEIKO Controller:
Multi-Contact Whole-Body Force Control for Position-Controlled Robots

Website: https://hucebot.github.io/seiko_controller_website
Video: https://youtu.be/RGkZS57_6Nk
Paper: https://ieeexplore.ieee.org/abstract/document/10517393
ArXiv: https://arxiv.org/abs/2312.16465
HAL: https://hal.univ-lorraine.fr/hal-04362547
Authors: Quentin Rouxel, Serena Ivaldi, Jean-Baptiste Mouret

This repository contains the implementation of SEIKO (Sequential Equilibrium Inverse Kinematic Optimization) associated with the paper Multi-Contact Whole-Body Force Control for Position-Controlled Robots, developed at INRIA Nancy, Larsen Team, in 2023 ; and the paper Multicontact Motion Retargeting Using Whole-Body Optimization of Full Kinematics and Sequential Force Equilibrium, developed at the University of Edinburgh in 2021. This is the code that was run on the Talos humanoid robot during multi-contact experiments. This repository does NOT include the implementation of various tests, simulation environment, viewer and human-interface used to teleoperate the robot.

Implementation Overview

• Implementation of SEIKO Retargeting QP, Cartesian input commands processing and contact switch procedure:

inria_model/include/inria_model/SEIKORetargeting.hpp
inria_model/src/SEIKORetargeting.cpp
inria_model/include/inria_model/SEIKOWrapper.hpp
inria_model/src/SEIKOWrapper.cpp

• Parameters of SEIKO Retargeting for Talos robot:

inria_model/include/inria_model/SEIKOTalos.hpp
inria_model/src/SEIKOTalos.cpp

• Implementation of SEIKO Controller QP:

inria_model/include/inria_model/SEIKOController.hpp
inria_model/src/SEIKOController.cpp

• Main controller pipeline thread running at 500 Hz:

inria_controller/include/inria_controller/TaskSEIKO.hpp
inria_controller/src/TaskSEIKO.cpp

• ROS_Control lowlevel controller running in position-control mode at 2 kHz and performing joint position commands interpolation and safety checks:

inria_controller/include/inria_controller/ControllerBase.hpp
inria_controller/src/ControllerBase.cpp
inria_controller/include/inria_controller/ControllerTalos.hpp
inria_controller/src/ControllerTalos.cpp

• Wrappers around RBDL and Pinocchio libraries to manipulate the robot's model:

inria_model/include/inria_model/Model.hpp
inria_model/src/Model.cpp
inria_model/include/inria_model/PinocchioInterface.hpp
inria_model/src/PinocchioInterface.cpp

• URDF model of the Talos robot with 3d-printed ball-shaped right hand effector:

inria_talos_description/urdf/talos_stump.urdf

• Starting script to load and run the ROS_Control controller on the robot:

inria_controller/scripts/startTalos.sh

Dependencies

This implementation is intended to be compiled with catkin under ROS Melodic. It requires the following catkin packages:

• RBDL (rigid body dynamics)
  https://gitlab.inria.fr/locolearn/rbdl_catkin
• Pinocchio (rigid body dynamics for analytical derivatives)
  https://github.com/stack-of-tasks/pinocchio
• eiquadprog (QP solver)
  https://github.com/stack-of-tasks/eiquadprog
• RhIO (user and command interface, monitoring, configuration, logging)
  https://gitlab.inria.fr/locolearn/rhio
• ZMQ (network transport for RhIO)
  https://gitlab.inria.fr/locolearn/libzmq_catkin

Citation

SEIKO Controller:

@article{rouxel2024multicontact,
  title={Multi-Contact Whole-Body Force Control for Position-Controlled Robots},
  author={Rouxel, Quentin and Ivaldi, Serena and Mouret, Jean-Baptiste},
  journal={IEEE Robotics and Automation Letters},
  year={2024},
  publisher={IEEE}
}

SEIKO Retargeting:

@article{rouxel2022multicontact,
    title={Multicontact motion retargeting using whole-body optimization of full kinematics and sequential force equilibrium},
    author={Rouxel, Quentin and Yuan, Kai and Wen, Ruoshi and Li, Zhibin},
    journal={IEEE/ASME Transactions on Mechatronics},
    volume={27},
    number={5},
    pages={4188--4198},
    year={2022},
    publisher={IEEE}
}

License

Licensed under the BSD License