Spring 2021 GRASP SFI: Zackory Erickson - April 7th, 2021
Zackory Erickson, a PhD candidate in Georgia Tech’s Healthcare Robotics Lab, discusses robotic caregiving and physical human-robot interaction with the GRASP Lab audience.
Zackory Erickson, an assistant professor at Carnegie Mellon University’s Robotics Institute, discusses capacitive sensing for physically assistive robots and caregiving with the HIRO Group.
Erickson’s lab studies robots that touch, sense, and assist people with daily care. Cameras can lose sight of a person when the robot or clothing blocks the view, while force sensing requires contact and pressure mats can miss local pose changes or motion in real time.
Capacitive sensors estimate distance to the body without requiring contact, and can work through materials such as cotton clothing. A single-axis sensor let a PR2 stay about five centimeters above participants’ arms as they moved during dressing; in a study of ten able-bodied participants, it also followed arm contours. Erickson described a separate case where the approach adapted to involuntary arm motion during gown dressing for a participant with cerebral palsy.
To capture motion beyond vertical distance, the group built a sensor with six copper electrodes. Its model estimates four aspects of relative limb pose: lateral and vertical distance, plus pitch and yaw. The sensor samples above 100 hertz and works from roughly 10 to 15 centimeters away; a neural network trained on one participant’s data achieved average position error below 2.5 centimeters and orientation error below 10 degrees across 12 participants.
Using a PR2 and a simple feedback controller, the team kept the end effector within about one centimeter of its target while tracing arms, elbows, legs, and knees. In dressing and bathing demonstrations, the robot adapted to participants’ movement and cleaned limbs through a wet washcloth. Wet cloth sometimes disrupted sensing, however, and the robot briefly lost track near a knee; Erickson also said limb size can contribute to estimation error because it changes capacitance.
Erickson said force-torque sensing can be useful for detecting contact or slipping, while capacitive sensing can help a robot follow limb contours and navigate around joints. He suggested combining vision, capacitive sensing, and force-torque sensing could offer complementary information for physical assistance.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.
Zackory Erickson, a PhD candidate in Georgia Tech’s Healthcare Robotics Lab, discusses robotic caregiving and physical human-robot interaction with the GRASP Lab audience.