Scientists develop a smart bionic robot

Home > Sci-Tech

Scientists develop a smart bionic robot

2022-01-15 18:03:44 15 ℃

Vision-sensitive disorders caused by buried accidents such as earthquakes, landslides, landslides, have hindered the search and positioning of the trapped personnel. In order to ensure the best rescue timing, improve search and rescue success rate and survival rate, relying on national technology innovation 2030 "new generation of artificial intelligence" major projects, the identification and rescue requirements of the trapped personnel in extreme environments without visual input, Chinese Academy of Sciences Shanghai The Tihu team, the research staff of the Microsystem and Information Technology Research Institute, inspired by the natural world star nose, "sniffing", combining the MEMS olfactory, tactile flexible sensor array with multi-modular machine learning algorithm, and constructing the mouse tuning of the mouse Sniffing intelligent robot (as shown). Thanks to the silicon-based MEMS gas sensor (sensitivity transcendence 1 quantity level), the excellent performance of the pressure sensor (detection limit transcendence humans), the robot finger touchs the object to accurately acquire its local micro-shape, material hardness and The overall contour, etc. , Closed loop rescue.

Shanghai Micro System Institute and the Emergency Management Department, Shanghai Fire Research Institute, through the field research on the first-line fire rescue unit, the real restore bury scene covered by the human body, in this environment, including the human body 11 Typical objects are identified, and the accuracy of the sniffing association has reached 96.9%, which is 15% more than a single feel. A single tactile (548 sensor) perceived by Nature (DOI: 10.1038 / S41586-019-1234-Z), the work is hit (70), sniff (6), Only 1/7 of the number of sensors use more ideal purpose, and the reduced sensor size and sample amount are more suitable for fast reactions and applications under limited conditions in resources. In addition, in the case of the presence of interference gas or device partial damage in actual rescue, the system maintains a good accuracy (> 80%) through multi-modal perceived complementary and fast adjustment of neural networks.

Related research results Take a Star-Nose-Like Tactile-Olfactory Bionic Sensing Array for Robust Object Recognition in Non-Visual Environments, published on Nature Communications on January 10, 2022. Research work has been supported by scientific and technological innovation 2030-major projects, national key research and development programs, the National Natural Science Foundation of the National Natural Science Foundation, and the Sino-Science Research Program, from 0 to 1 "original innovation project, Shanghai Excellent academic leader program and so on.

Sniffing a bionic intelligent robot for emergency rescue buried scene