HUMAN ROBOTICS

 modeling of control in humans for

collaborative robots and haptic interfaces

This is the official webpage of the tutorial on human robotics that will be held on September 29, 2004 at the IEEE/RSJ International Conference on Intelligent Robots and Systems (http://www.iros2004.org/)

Experiments using a haptic interface to investigate how humans interact with novel dynamic tasks (Left, Nature 414: 446-9, 2001) and scheme of modeling of neuromechanical control in humans, which can be used to control collaborative robots.

Date and Venue

This tutorial will take place Wednesday September 29 2004 from 09:45 to 12:45 and 13:45 to 16:45 in meeting room 6 of the Sendai international centre, Sendai Japan

How to register?

To sign up for the tutorial, please mark it during the online registration procedure or send an e-mail to the organizers of IROS2004

Targeted Audience
This course is for roboticists developing collaborative robots, and for anyone interested in understanding the close relationship between robot control and human motor control. Engineers working or who intend to work in the emerging field of neurobotics, as well as psychologists who want to learn robotic modeling of human motor control may also find it particularly useful.

Motivation and Objectives
    The last few years have seen considerable research on collaborative robots for manufacturing and automotive industries, humanoid robots, robotic assistive devices for medical interventions, virtual reality based training using haptic interfaces, and on the use of robotics in rehabilitation and in neural engineering. In some cases, the efficiency of these novel techniques has already been demonstrated. For example, impressive robots have been developed for minimally invasive surgery. However it is generally recognized that the human-machine interface requires significant improvements.

Precise knowledge of the mechanisms involved in motor control and adaptation is critical to all of these applications and may result in improved human-machine interaction. What is required are models of human motor control that  i) reproduce well the dynamic behavior of limbs as observed in experiments, and  ii) can be implemented on robotic systems and haptic interfaces. Such models have been developed by analyzing the biomechanics and neural control of human motions and by using nonlinear adaptive control and machine learning techniques for modeling (http://www.bioeng.nus.edu.sg/research/HumanROBOTICS/human-main.htm).

    This tutorial will present computational models of neural control based on measurable variables. It will provide knowledge about the biological actuators (muscles), the sensors (e.g., Golgi tendon organs and muscle spindles), and the dynamics of the musculoskeletal system and brain activity, gained from data acquired using camera-based systems, haptic interfaces, electromyography (EMG) and Functional Magnetic Resonance Imaging (fMRI). The results of a number of experimental studies are used to guide robotic modeling of the neuromechanical control in humans. Such robotic controllers can be used for nonlinear adaptive control of (collaborative) robots and haptic interfaces. They can also be used to simulate the effect of neuromuscular disorders on control as well as to develop better controllers for neural prostheses and robot assisted rehabilitation protocols.

The participants will be given comprehensive tutorial notes with references which can be used for the tutorial and further study.

List of topics to be covered

· muscle mechanics and simple reflexes

· mechanical properties of joints and limbs, mechanical impedance of coupled systems, and control of limb mechanics

· integration of properties of the biological sensors and the dynamics of the musculoskeletal system, towards human robotics

· nonlinear control (e.g. impedance and adaptive control) applied to modeling of human motor control; this includes motor adaptation, a fundamental feature of human motor behavior

· neural control of manipulation, i.e. actions performed in interaction with the environment

· application of control models to simulate disordered neuromuscular control and to development of better controllers for neural prostheses and robot assisted rehabilitation protocols

About the Speakers

Ted E Milner is an expert in human biomechanics and neural control and is currently developing robotic systems and haptic interfaces for assessment of motor function and rehabilitation. Etienne Burdet is doing research in robotics and human-machine interaction, and has made significant contributions in both of these fields. The recent results they have obtained together contributed to significantly clarify how the central nervous system controls muscles to perform stable and unstable tasks (Nature 414: 446-9, 2001), and led to models of how the central nervous system (CNS) adapts to novel stable or unstable dynamics, which can be used for robot control.

Ted E MILNER School of Kinesiology Simon Fraser University Canada tmilner @ sfu.ca http: //css.sfu.ca/sites/ncl/

Etienne BURDET Dept. of Mechanical Engineering and Division of Bioengineering National University of Singapore e.burdet@ieee.org http://guppy.mpe.nus.edu.sg/~eburdet/