Soft robots are now becoming better at feeling the human touch. Thanks to Cornell University researchers. They have developed soft robots that could detect any human physical interaction without any form of touch. This is capable due to a USB camera and machine learning software.
- The USB camera is placed inside the robot and it captures the shadow of the hand gestures made on the skin of the robot.
- While the machine learning software classifies it.
The Shadow sense technology is the project from the Human-Robot Collaboration and Companionship Lab led by Guy Hoffman an Associate Professor of Sibley School of Mechanical and Aerospace Engineering.
Instead of using contact sensors in the robot which could cause more weight and difficult wiring, the team made use of the sight to test the touch. Hu, the lead author of the group paper ShadowSense: Detecting Human Touch in a Social Robot Using Shadow Image Classification, said that: "the camera inside the robot can understand how the person is touching and what is the intention of the person by looking at the shadow images."
Features of the Robot
- It consists of a soft bladder of nylon skin which is inflatable. This is stretched around a skeleton cylindrical in shape. Its height is about four feet and is placed on the mobile base.
- A USB camera is placed under the robot's skin and is connected to a laptop.
- A neural network algorithm is developed that makes use of the recorded data to understand the basic gestures.
- During emergency evacuations, the robot could guide people to safety by communication.
- No use of contact sensors.
- An accuracy of 87.5 to 96% is achieved based on the lightning.
- Could be programmed to react to gestures and touches.
- As an interactive screen, the robot's skin could be used.
- By training the robots, they could understand more interactions.