Several groups are developing touch electronic skins for robots. However, Cornell University scientists are pursuing a simpler approach, using a camera that captures darkness to give a robot sensation.
Known as ShadowSense, the test system incorporates a regular camera connected to a USB-enabled laptop, located below are not– Electronic translucent “leather” on a soft body robot.
As a person approaches the robot, the ambient light will cast the shadow of their hand on the skin. The camera tracks that ball from the other side of the skin (in the robot), using machine learning-based algorithms to determine when the hand actually touched the skin, the area where it touched, and its gestures. In this way, ShadowSense can not only tell when and where to touch the robot, but also can assign different commands to different touch gestures.
The current robot prototype – essentially just a nylon leather inflatable bladder stretched around a cylindrical wheel skeleton – capable of distinguishing between palm touch, punching, two-hand touch, hug, thread and don’t touch. It can do so with an accuracy of between 87.5 and 96 percent, depending on the intensity and direction of the light.
Researchers were quick to point out that the technology’s applications are not limited to robots, as it can also be used in touch screens or electronic devices. That said, the current ShadowSense still has some limitations – not just the light source, but the camera also has to be placed within view of the interactive part of the skin. Using additional mirrors or lenses can help solve the following problem.
“Induction is an important means of communication for most organisms, but it is virtually absent from human-robot interactions,” said lead scientist, Assoc. Professor Guy Hoffman. “One of the reasons is that full-body induction is used to require a large number of sensors and is therefore impractical to do. This study provides a low-cost alternative.”
ShadowSense is illustrated in the video below and described in an article recently published in the magazine The Computing Association’s Proceedings for Interactive, Mobile, Wearable and Universal Technologies.
Source: Cornell University
Researchers let soft robots touch humans