Exo Motor Skills Learning

EAGER: Principles of Motor Skills (re)Learning When Using Powered Exoskeletons


Exoskeletons have great potential to augment physical power while preserving human skill in dynamic industrial environments. However, many industrial tasks involving tools are complex, and require high dexterous control. Workers also have different levels of experience and possess different physical and cognitive skill-sets. Hence, exoskeleton acceptability and appropriate training are key factors that will determine future industrial applications. This project will promote the progress of science and advance the national health, prosperity and welfare by identifying basic principles of human motor skill learning when using multi-joint arm exoskeletons to perform complex tasks. By introducing exoskeletons to users with varying levels of task-relevant skills at baseline, and by modeling learning using a comprehensive set of neuromotor measures, the project promises to discover fundamental principles that will guide the design of adaptable exoskeleton control algorithms and personalized training protocols that can produce true collaboration between exoskeletons and individuals with diverse skills. Broader impacts of the work include workshops focused on increasing workers awareness of exoskeletons and their potential applications in the workplace, as well as opportunities for worker-skills training using exoskeletons.
This work will advance knowledge of the complex learning dynamics that arise in human-exoskeleton systems through systematic manipulation of critical mind-motor-machine factors influencing the collaboration between man and machine. Our novel contribution will be to expand the generalizability of a two-stage multi-rate model of motor adaptation by studying short- and long-term adaptations with exoskeleton use by a broad range of users with varying skill and tasks of varying complexity. The concept of persistent excitation will be leveraged to perturb the dynamical conditions under which the human accomplishes the task, to maximize human learning from the machine, while an extremum seeking control approach will be implemented for the machine to learn from the human response. The two concepts will be combined in a novel exoskeleton control algorithm for adaptively tuning exoskeleton parameters (such as torque amplification) as the humans use of the exoskeleton evolves.


Alexander Leonessa

Professor, Mechanical Engineering Department

Virginia Tech, Blacksburg, VA 24060

Divya Srinivasan

Professor, Industrial Engineering and Bioengineering

Clemson University, Clemson, SC 29634