Jesse Weisberg is a Washington University alumnus with a Masters of Engineering in Robotics and is the project leader of Therabotics. Therabotics’ vision is to improve the control of robotic prosthetics and orthotics, starting with the prosthetic hand. He and his teammates, Pin-Wei Chen and Sri Harsha Kondapalli, created a series of functional prototypes that present how computer vision can be used to make a prosthetic hand that is both easier to use and has more functionality than ever before. Therabotics combines eye-tracking technology with a sensor-enriched artificial intelligence system, so that the robotic device becomes aware of the user’s intended grasp. Jesse won the Sling Health summer fellowship in which he received a $5,000 stipend to work at CET (an innovation center that houses many Sling Health alumni) to further develop the technology that his team created during the academic year.
Here is a Q & A with Jesse about his experience:
Q. Why did you apply to the Sling Health Fellowship program?
A. I applied to the Sling Health Fellowship program because I was working on an idea that I was very passionate and excited about, and wanted to further my work. Bringing this idea to fruition, even at the prototype stage, was going to require full-time dedication for a number of months. I was excited to fully immerse myself in this project, and with the resources from Sling Health, I felt it was very possible.
Jesse Weisberg (left), MS in Robotics and Pin-Wei Chen (right), PhD in Rehabilitation and Participation Science
Q. What problem were you trying to solve?
A. We found that robotic hand orthoses/prostheses are simply too difficult to use, and prevents most patients who need such a device to use one. Over 3 million people suffer from upper limb paralysis and need these robotic assistive devices. However, roughly 65% of this population does not use these devices. Rehabilitation researchers and a number of clinical rehabilitation specialists agree this is because existing solutions are too difficult to use. From an engineering standpoint, we discovered the root of the issue stemmed from limitations in myoelectric control. We set out to create an alternative control paradigm that fused useful sensory information additional to myoelectric control to make a more robust and intuitive robotic prosthetic hand.
Sri Harsha Kondapalli, PhD in Electrical & Systems Engineering
Presentation on 2017 Demo Day
Q. What is the future of Therabotics?
A. We are continuing to iterate through prototypes to make the system more practical, functional, and reliable. Technically, we are currently working on moving from using real-time object detection to a reinforcement learning architecture. Such an architecture will allow grasps and movements of the prosthetic hand to be learned which each use, so that the functionality of the prosthetic adapts to the user’s movement over time. Through this effort, our hope is to build a robotic prosthetic hand that can achieve truly fine control.
You can follow Jesse and Therabotics at this website: https://www.jesseweisberg.com/therabotics/
Demonstration of the Object-Specific Grasps with Eye Tracking by Jesse Weisberg