Mobile Therapy Assistants

The development of mobile therapy assistants is a main focus of our lab. Why, you ask?  In developed countries, lower birth rates and an aging population are leading to a shift towards more senior citizens per working-age adult. This is creating a healthcare crisis in which there are not enough rehabilition professionals to meet the needs of society. For example, Japan’s ratio of seniors to adults aged 15-64 will reach 1:1 by 2025. China’s one-child policy will lead to a single child taking care of two parents and four grandparents. While the numbers aren’t as alarming for the U.S. and Europe as they are for various Asian countries, the trend is the same. Conversely, in lower income countries, labor might be cheap and they do not face the same demographic trends, but inadequate infrastructure has failed to train the requisite numbers of rehabilitation professionals to meet the needs of their societies. It is for these reasons, that we are working on the below projects.

Improving Telerehabilitation in Pediatric Cerebral Palsy Using Machine Learning and Social Robots

Limits to care for pediatric cerebral palsy patients in rural areas could be addressed through telepresence, but current technologies fail to adequately convey information. This project will explore whether combining social robots with telepresence systems and computer vision can improve the quality of telerehabilitation for the purpose of remote assessment. The social robot will aid communication to the patient, motivating them through activities, while the computer vision automatically assesses function.

Here is a video overview of our system:

Mobile Service Robot for Task-Oriented Stroke Therapy



  • Grant: The LaVerne L Schwacher Endowed Fund for Stroke Research
  • Department of Physical Medicine and Rehabilitation, Medical College of Wisconsin
  • Daren Hughes (MCW), Rachel Wilk (Marquette University)
  • Department of Physical Medicine and Rehabilitation, University of Pennsylvania, School of Medicine
  • GRASP LAB, , University of Pennsylvania, School of Engineering and Applied Science
  • Jennifer Dekerlegand, Good Sheppard Penn Partners

Robot-Assisted Stroke Rehabilitation Based on Patient Therapy Interactions (BAXTER)

Baxter, a humanoid robot, is a perfect candidate to act as a therapist to stroke survivors. The current study involving the Baxter robot involves three stages:  learning, demonstration, and teaching.  In the learning phase, Baxter is trained how to mimic a real-life therapist;  this is done by using Microsoft Kinect, which maps the kinematic configuration of the human to Baxter.  After the robot demonstrates what it has learned from the therapist, the robot ostensibly would be capable of giving the desired therapy to the patient.  If you notice, the manipulators on the Baxter robot are not appropriate for assisting human movements, hence robotic grippers with precise tactile sensing are being developed in the lab which would make the robot capable of performing dextrous manipulation on a human being.


  • Grants:  NIH K25 Grant #1K25NS058577 – 01A1
  • Rehabilitation Robotics Research and Design Lab
  • Department of Penn Medicine and Rehabilitation, School of Medicine
  • GRASP LAB, , University of Pennsylvania, School of Engineering and Applied Science
  • Jennifer Dekerlegand, PT, Good Sheppard Penn Partners