Visual Servoing for Flexible Robot Manipulators

A method to solve the inverse kinematics problem for robot manipulators with flexible links

This project is done as our class project for CMU 16-711 Kinematics, Dynamics and Control.

Authors

• Haowen (Harvey) Shi (haowensh@andrew.cmu.edu)
• Daqian Cheng (daqianc@andrew.cmu.edu)
• Hengrui (Henry) Zhang (hengruiz@andrew.cmu.edu)

Abstract

Generally, to solve the Inverse Kinematics (IK) problem of a rigid manipulator, we first need its kinematic parameters (D-H or twists) obtained via the mechanical specification and/or external calibration. However, flexible robots have changing parameters and hence their IK cannot be solved in this way.

In this project, we proposed a method to solve the IK problem for the robot manipulators with flexible links. We formulate the problem as an IK problem for manipulators with unknown geometry, and iteratively & numerically estimate the twists (kinematic parameters) of our manipulator on the fly. Our pipeline consists of two major parts: an AprilTag based end-effector pose estimation and iterative IK solver.

Framework

Results

This video shows our experiment with a 9 DoF snake-like arm with flexible links (modelled using friction joints on top of motorized joints) in CoppeliaSim simulation environment.

For more information

View our project code on GitHub:

jacobian-visual-servo

Download our project report