Sensorless Transparency-Optimized Control Architecture based on Inverse Dynamics Modeling for the WAM Teleoperation System

Amir Noohian, Dylan Miller, Justin Valentine, Alan Lynch, and Martin Jagersand

WAM bilateral teleoperation system setup with a 4-DOF leader arm equipped with a custom haptic wrist (left) and a 7-DOF follower arm (right).

Overview

Human-scale bilateral teleoperation allows operators to interact with remote environments through robotic manipulators, but achieving natural and intuitive interaction remains challenging due to the robots’ size and dynamics. The goal of this work is to improve transparency in human-scale teleoperation, enabling operators to perceive the remote environment as if directly manipulating it.

Abstract

The four-channel teleoperation architecture is a well-established framework for achieving transparency in bilateral robotic systems; however, attaining high transparency in human-scale teleoperation remains challenging due to large robot inertia, nonlinear dynamics, and the practical limitations of force/torque sensors. This work presents a sensorless four-channel teleoperation approach that leverages inverse dynamics modeling to provide dynamic compensation and estimate external interaction forces without dedicated sensors. The proposed system is implemented and experimentally validated on a WAM bilateral teleoperation system composed of two Barrett WAM manipulators. Experimental results in free-motion and contact-rich tasks demonstrate improved position tracking, more accurate force reflection, reduced operator effort, and higher transparency compared to conventional as well as transparency-enhancement teleoperation methods.

Sensorless Four-Channel Control Architecture on the WAM Teleoperation System

This video presents an overview of the complete study, highlighting the proposed human-scale bilateral teleoperation framework and its experimental validation on a dual Barrett WAM platform. It illustrates the system architecture, control concept, and key results in both free-motion and contact-rich tasks, demonstrating improved transparency and intuitive force feedback.

Control Architecture

The control architecture follows a four-channel bilateral teleoperation structure, consisting of position and force channels for both the leader and follower manipulators. Model-based dynamic compensation is incorporated to counteract inertial, Coriolis, and gravitational effects, reducing the apparent impedance of the system. External interaction torques are estimated using the identified robot dynamics, enabling sensorless force feedback and improved transparency during contact interactions.

Evaluation

The objective evaluation framework compares transparency performance across different teleoperation systems using free-motion position tracking accuracy, leader impedance, hard-contact force tracking, and maximum transmittable impedance. Overall, the proposed 4c-DynComp controller consistently outperforms all other methods, achieving the best free-motion tracking, the lowest leader impedance, and the highest maximum transmittable impedance (𝑝<0.05), while maintaining accurate force tracking during contact interactions.

Whole-Body Manipulation

The system enables whole-body interaction, allowing the follower robot to use both its end-effector and other parts of the arm during contact tasks. In the door-opening demonstration, the robot manipulates the handle and then uses whole-arm contact to push the door open.

Territorial Acknowledgement

The University of Alberta, its buildings, labs, and research stations are primarily located on the traditional territory of Cree, Blackfoot, Métis, Nakota Sioux, Iroquois, Dene, and Ojibway/Saulteaux/Anishinaabe nations; lands that are now known as part of Treaties 6, 7, and 8 and homeland of the Métis. The University of Alberta respects the sovereignty, lands, histories, languages, knowledge systems, and cultures of First Nations, Métis and Inuit nations.

Page updated

Report abuse