Quadruped manipulator framework

This repository provides all necessary resources to replicate a custom 3-DoF robotic arm designed for legged robots. The arm is built with a focus on being lightweight, easy to maintain, and capable of handling payloads without compromising functionality.

The framework is open-source and includes a script-like language for sending high-level interpolation commands, similar to commercial modules. The entire implementation is written in C++ and has been tested in ROS Melodic and Noetic.

Packages/Nodes

• quadruped_igus_bringup - launches the complete system, including both simulation and hardware setups;
• quadruped_igus_description - contains everything related to the URDF, including a launch file to display the robot model in RViz with the joint_state_publisher GUI for workspace visualization.
  • Robot model obtained from: Unitree Robotics GitHub
  • The arm model was created using Fusion 360 and then converted to URDF with fusion2urdf.
• aliengo_controller - quadruped controller adepted from: A1 Simulation Python Repository
• igus_arm_driver - robotic arm controller implementing an IK-based motion planner with two movement interpolation types: MOVJ (joint space) and MOVL (linear). This package also includes the firmware for communication with IGUS Rebel joints.
• igus_rviz_panel - custom RViz panel to facilitate sending arm movement instructions directly from the visualization interface

Prequesites

• trac_ik_lib - sudo apt install ros-noetic-trac-ik-lib

  If you encounter the following error while compiling:

  /opt/ros/noetic/include/trac_ik/nlopt_ik.hpp:35:10: fatal error: nlopt.hpp: No such file or directory
    35 | #include <nlopt.hpp>

  Install the missing dependency with: sudo apt install libnlopt-cxx-dev

• effort_controllers (for simulation use only) - sudo apt install ros-noetic-effort-controllers

Usage

Simulation

1. Run the environment:

roslaunch quadruped_igus_bringup run_simulation.launch

2. Unpause Gazebo to start the system.

Hardware

1. Change the specific information about the arm in src/igus_arm_driver/config/arm_setup.yaml file.
2. Run the environment:

roslaunch quadruped_igus_bringup run_robot.launch

3. System is ready when the terminal says "All joints disabled".

Send arm instructions

There are two ways of sending instructions to the arm:

• RViz Panel
  Simply fill in the fields and press Go.

  

• Publish a goal in igus_arm_driver_skill/goal target space
  Send one movement or a sequence of movements using instructions of the following type:

  moveT(p={a,b,c},v=s,a= $\ddot{\theta}_{max}$ )

  Here's what each part of the command means:

  • T: Interpolation type. This defines how the movement should be performed. The available types are:

    • j: Joint space interpolation (goal defined in 3D space)
    • jj: Joint space interpolation (goal defined in joint space)
    • l: linear interpolation (goal defined in 3D space)
    • lj: linear interpolation (goal defined in joint space)

  • (a,b,c): The goal for the movement.

  • s: The speed for the movement.

    • For joint goal movements (jj or lj), s is the maximum joint speed.
    • For 3D goal movements (j or l), s is the tool speed.

  • $\ddot{\theta}_{max}$: The maximum joint acceleration for the movement.

    A program ends when ; is found, each instruction must be separated by a comma.

  Example: 4 sequenced joint-space movements:

  movejj(p={0.79, 1.87, 2.04},v=0.6,a=0.7,b=0),movejj(p={2.53,2.31, 1.43},v=0.6,a=0.7,b=0),movejj(p={0.55, 2.31, 1.43},v=0.6,a=0.7,b=0);
  

Limits

• Maximum TCP velocity: 0.2 m/s
• Maximum joint velocity (w_max): 0.6 rad/s
• Maximum joint acceleration (a_max): 0.7 rad/s²

All these values can be modified in src/igus_arm_driver/config/arm_setup.yaml. Additional configurable parameters include:

• w_min: The deadzone limit for joint velocity.
• error_threshold_joints: The acceptable stopping error in joint space.
• error_threshold: The acceptable task space error, measured in meters using Euclidean distance.

These parameters influence motion control accuracy and can be adjusted based on system requirements. The kinematic limits for IK are defined in src/igus_arm_driver/src/igus_arm_driver/trac_kinematics.cpp in order to keep the solutions inside the desired workspace.

License

Distributed under the MIT License. See LICENSE for more information.

Contacts

If you have any questions or you want to know more about the work developed by us, please contact one of the contributors of this project:

• Maria S. Lopes(email:inesctec, email:feup, orcid) (corresponding author)
• Alexandre Melo (email, orcid)
• João Pedro C. de Souza (email, orcid)
• Pedro Costa (email, orcid)
• Manuel F. Silva (email, orcid)

Acknowledgements

• CRIIS - Centre for Robotics in Industry and Intelligent Systems from INESC TEC - Institute for Systems and Computer Engineering, Technology and Science
• Faculty of Engineering, University of Porto (FEUP)
• ISEP - School of Engineering of the Porto Polytechnic

Funding

This work is financed by National Funds through the Portuguese funding agency, FCT - Fundação para a Ciência e a Tecnologia, within project LA/P/0063/2020. DOI 10.54499/LA/P/0063/2020