ROS2 Environment for the simulation and control of a tendon actuated robot using PyElastica

[gabotuzl]

Hello!

I wanted to share my work, given that PyElastica was used as the simulation software in my project.

My project consists of two ROS2 networks, which can carry out open-loop and closed-loop simulations of a tendon-actuated continuum robot.

In general, the operation of the open-loop system, is as follows:

• First, the simulation, visualization, and open-loop graphical user interface nodes are powered on.
• In the graphical user interface, all parameters and conditions desired for the simulation are specified, including all parameters and voltages for the tendons that the user wishes to add.
• The data is sent, which is received simultaneously by the simulation and visualization nodes.
• The simulation node initiates the simulation, and when it is finished, it sends the position and orientation history data of the rod nodes to the visualization node.
• The visualization node is responsible for creating a video that shows the evolution of the system over time.

In general, the operation of the closed-loop system, is as follows:

• First, the simulation, visualization, controller, gain searcher, and graphical user interface nodes for control are powered on.
• In the graphical user interface, all parameters and conditions desired for the simulation are specified, including those related to the control of the robot (desired XYZ position of the robot tip's position). Additionally, it is specified whether or not to perform a proportional gain search (given that proportional control is being used, gains must be specified).
• The data is sent, which is received simultaneously by different nodes.
• The simulation node begins the simulation and simultaneously sends position data to the controller node, which responds with tensions for the robot's tendons, attempting to reach the desired position with the given gains.
• Once the simulation time is complete, the simulation node publishes the position history of all the nodes in the robot to the visualization node, which is responsible for creating a video that shows the evolution of the system over time.
• If the gain search option is selected, the gain search node will publish new gains based on the obtained results to the controller node, and a new simulation will begin in the simulation node. This process continues until the time convergence criteria is met.

The purpose of this ROS2 environment is to facilitate the simulation and implementation of control schemes to a tendon-actuated robot system. The ROS2 architecture allows for anyone to add or remove nodes to change the function and/or add more utility to the network. Using PyElastica allows for the user to accurately simulate the complex dynamics of a flexible rod robot, all the while maintaining the computational cost at a low. Although this project deals with tendon-actuated robots, any type of external forcing can be added.

This project's repo is the following: https://github.com/gabotuzl/tendon_cr_environment

Please, if you have any suggestions, comments, or questions, send them my way!