Analysis of body-fin cooperative propulsion mechanism of koi carp using Pose Estimation^⋆

ABSTRACT Advances in science and technology have led to the development of advanced underwater robots, significantly improving humanity’s ability to understand, explore, exploit, and protect the oceans. The secret of efficient swimming in fish provides an essential theoretical basis for developing new underwater robots with low noise, high efficiency, high stability, and high maneuverability. The tools or methods to study the movement of fish are the key to unlocking the secret of efficient fish swimming. Therefore, this study builds an pose and motion estimation platform based on advanced computer vision technology for studying fish motion. The platform can quantify the motion of fish, enable 3D pose estimation of the fish body, and dynamically visualize the motion data of marked vital points. We used this experimental system to study the movement of koi (forward, backward, floating, and turning, etc.) to demonstrate the advantages of the experimental system we designed. The research results provide innovative ideas for fish motion research methods, which can be valuable in designing, developing, and optimizing modern underwater robots. The code and dataset of this study are available at https://github.com/wux024/AdamPosePlug . Graphical Abstract Highlights A novel pose estimation-based experimental system for representing fish swimming is designed. A software that integrates experiments on fish swimming studying is designed. An insight into the synergistic of the fins of koi in four swimming states is proposed.