Endow underwater robots with water flow sensing intelligence, Chinese scientists achieve new breakthrough in biomimetic perception.

In nature, fish rely on the lateral line system to perceive changes in water flow, allowing them to navigate and avoid obstacles in turbulent waters. Inspired by this, the Intelligent Robot System Research Team at the Chinese Academy of Sciences Institute of Automation has innovatively developed a bio-inspired lateral line sensor called FlowSight, based on autonomous vision. This sensor gives underwater robots precise "water flow sensing wisdom", opening up a new path for autonomous navigation and environmental monitoring tasks in complex water environments. This bio-inspired lateral line sensor uses a flexible silicone feeler resembling a fish lateral line neuromast to sense water flow. When the feeler is struck by water flow, its deformation information is captured in real time by a built-in high-definition camera as an image sequence. Utilizing deep learning models, the sensor analyzes the water flow speed and direction from the images, achieving efficient vector sensing of water flow without the need for additional equipment. The relative error in water flow speed measurement is 3.05% and the relative error in water flow direction measurement is 0.98%. Furthermore, the research team has integrated this sensor into a bio-inspired underwater robot and successfully conducted closed-loop motion control experiments based on water flow sensing. The underwater robot can swim against the current and dynamically adjust its posture, similar to real fish, providing a new solution for underwater exploration and ecological monitoring.