ERI: Rapid-Manufactured Knitted Sensing Skins for Assistive Robots

Wesleyan University, Middletown CT

Investigators

Abstract

While human skin makes it possible for us to detect low-force contacts with our environments anywhere on our bodies, robots can only feel contacts where their designers place sensors. Full-body sensitivity is especially important for robots that physically interact with people, who may contact the robot in unexpected locations or at unexpected times. In 2014, 24.4 million adults in the United States required assistance with activities of daily living such as feeding, bathing, or ambulation. Robots with sufficient power to perform assistive tasks such as helping a person transfer between a bed and a chair can physically injure a person through an unexpected contact. If a robot can detect these contacts through a sensitive skin covering its body, it can stop the motion or move away – significantly increasing the safety and comfort of the user. This research has significant potential to positively transform society by reducing the risk of injury for the patient care workforce, creating more affordable options for in-home assistance, and supplementing the home care workforce. Existing pressure and shear force sensors are complicated to build and repair, and thus very expensive. Computational machine knitting provides an opportunity to create sensing skins of arbitrary shape and dimension. Machine knitting is a fast-manufacturing process for which build instructions can be automatically generated, meaning that sensing skins can be quickly produced with arbitrary sensor placement. Dr. Roberts has previously demonstrated a directionally selective knitted strain sensor made from inexpensive off-the-shelf parts: Silicone tubing, silicone cord, and conductive thread. The materials are also reusable and repairable, so the same materials can be used to make multiple skins for the same robot for different applications. This project will investigate how pressure and strain sensors for robot skins can be knitted automatically and at lower cost than state-of-the-art sensors. Objectives of this proposal include: 1) Developing capacitive knitted pressure sensors; and 2) developing directionally selective knitted shear strain sensors, with a computational framework to automatically generate patterns for sensors oriented in arbitrary directions. Sensors will be knitted in a sleeve for an assistive robot arm already being used for tasks such as bed-bathing and limb repositioning, and evaluated using contacts with known force profiles. In combination, these two types of sensors will be able to provide rich information about the contacts experienced by a robot as it interacts with people. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.