An Artificial Lateral Line-Based Active Obstacle Recognition Strategy and Performance Evaluation for Bionic Underwater Robots

In the dark or muddy environments of the ocean, the artificial lateral lines (ALLs) based on bionics are mostly used to detect oscillating obstacles, such as dipole sources. To further expand the function of ALLs, in this article, the active recognition of nonoscillating obstacles is realized by an ALLs based on a pressure sensor array. First, the experimental and simulation platforms with the sensor array are built, and a perception framework is proposed for processing the sensor data. Then, the simulation platform is verified through the pool experiment, and the mechanisms of pressure changes generated by static obstacles in the process of robot active recognition are analyzed. Finally, according to these mechanisms, the obstacles are recognized through a multilayer perception framework that considers the time and spatial context. The recognition results show that the ALLs and the perception framework are effective in the active obstacle recognition of robots. The effects of feature correlation, data volume, and sensor layout are also analyzed. In addition, the data from the pool experiment can obtain satisfactory perception results.