Human and Machine Learning for Customized Control of Assistive Robots

This work proposes to develop and evaluate a control approach that aims to make assistive devices more accessible to patients with severe paralysis. The approach will enhance the movement capabilities that survived the injury, through a combination of customized interfacing, machine learning, and human motor learning. Key components will include: (i) the use of a Body-Machine Interface to gather control signals from the user, building on ongoing work in the laboratory of PI Mussa-Ivaldi and (ii) the introduction of machine automation that furthermore is adaptable with machine learning techniques. Development of the Body-Machine Interface and testing will occur at the Robotics Laboratory, co- directed by PI Mussa-Ivaldi and located at RIC. Development of the robot autonomy software and hardware testing will occur at the Assistive & Rehabilitation Robotics Laboratory directed by PI Argall and located at RIC. The integrated system will be tested via user studies, with Spinal Cord Injury patients and uninjured subjects. All studies will be performed at RIC. IRB approval will be obtained from Northwestern for these studies. The goal will be to assess the Body-Machine Interface as a device to control various assistive robots and compare the learning outcomes induced in disabled users. No work will be performed on the Northwestern campus. The students funded on this project will be graduate students at Northwestern.