Underwater vehicle gesture control aided by air bubble motion observation

This work describes the gesture control of a Biomimetic Autonomous Underwater Vehicle (BAUV) in a water flow by utilizing information derived from an onboard stereo camera, a compass, and an accelerometer. In an alternating water flow, the BAUV suffers from drag forces and consumes more energy when it advances. The relationship between air bubbles and water flow is first discussed. The air bubble is detected by the Harris corner. The relative position between air bubble and BAUV is estimated based on the calibrated stereo camera and the bubble is tracked by Lucas-Kanade method combined with the image pyramid algorithm. By integrating observation information from the motion of air bubbles, heading angles and 3-axis accelerations, the BAUV adjusts its heading angle to optimize the gesture in the water flow by gaining lift forces from the flow. Finally, the gesture control aided by the bubble motion observation in a water flow is verified by experiments. The control energy consumed by the driving motor are calculated to compare the energy used in a water flow without the gesture control.