Design and development of new or improved rehabilitative and assistive devices to improve function for individuals with spinal cord injury or disease
Intelligent Assistive Robot Arm
Henny Admoni, PhD
Carnegie Mellon University
Pittsburgh, PA

Wheelchair-mounted assistive robot arms provide a flexible, mobile, highly dexterous tool that enables people with upper motor impairments to eat, drink, and care for themselves independently. However, controlling a robot arm can be difficult for people with motor impairments, who may only be able to use limited input devices (like a joystick or a sip-n-puff) to operate the robot. This project develops a robot arm that uses people’s natural, subconscious nonverbal behaviors to help predict what kind of assistance they need. Based on eye gaze and other behaviors, the system infers a user’s intentions and when they’re having difficulty with a task. State-of-the-art shared autonomy algorithms to can then use these predictions provide seamless robot assistance for completing tasks more quickly and with less effort. This project may help transform the way people use assistive robots, facilitating their re-entry into mainstream society and increasing their quality of life.