Today’s robots can recognize many objects by sight and touch using sensors and machine learning. However, they struggle with similar or unfamiliar objects and background noise. Researchers at Tsinghua University have developed a robotic touch sensing method that uses thermal sensing to improve object detection.
Their method uses spatiotemporal tactile sensing to detect thermal conductivity, diffusivity, surface roughness, contact pressure and temperature. They developed a layered sensor and an efficient classification algorithm for object detection.
Testing involved sorting garbage with a robot that achieved 98.85% classification accuracy. This system can reduce human labor and has wide applications for smart technologies. Future research will improve the intelligence of robots and explore combining sensors with brain-computer interfaces to help, for example, people without hands.
Source: Science Daily
