Embodied Sensing and Control in Fish-like Swimming Robots via Passive Degrees of Freedom

点击翻译按钮获取中文摘要

This Dynamics, Controls, and System Diagnostics award will support fundamental research into the analysis and design of fish-like swimming robots, enabling the use of passively articulated body segments to sense ambient flow characteristics and respond to hydrodynamic stimuli. This type of embodied sensing and control of locomotion is seen in swimming animals like fish, where sensory awareness of velocity and acceleration is used as feedback for control. The project will also incorporate active and passive internal rotors to extend beyond more familiar bio-mimetic approaches. The project incorporates ideas from several disciplines and combines theoretical models, numerical simulations, and experiments to translate these ideas to mechanical systems. The new capabilities for swimming robots will be particularly significant for power-limited underwater robots in sensory-deprived environments. The results of the project will enable the design of improved underwater robots for exploration, search-and-rescue, and inspection and maintenance of underwater infrastructure. The project will provide research training and career preparation to doctoral students, offer research and educational opportunities for undergraduate students through design projects, and will reach out to K-12 school students to attract them to STEM disciplines.This project seeks to investigate the role of passive degrees of freedom (PDOF) in aquatic robots that can act as both sensors and controllers. The dynamics of PDOF can encode the hydrodynamic forces and thus act as sensors. At the same time the motion of PDOF can be influenced by the hydrodynamic forces; if such motion is harnessed by appropriately designed PDOF, they can act as controllers. The project will improve our understanding of how the swimming systems state in a potential well encodes information about gaits and propulsive efficiency and how transitions between potential wells can be triggered by environmental disturbances or inputs that lead to fast and agile motion. The project will use machine learning methods to understand how the kinematics of a swimmer encode hydrodynamic information and how this can be used as part of multimodal sensory suite. Such embodied sensing and control would greatly reduce or in some cases bypass the role of electronic information processing and computation. The concepts of embodied sensing in aquatic robots can be applied to existing designs of robots by augmenting existing sensors for better sensor fusion algorithms. The modeling framework will also directly be applicable to several classes of terrestrial mobile robotic systems, where passive self-stabilization and agility could be improved by employing PDOF for embodied sensing and control. The specific research goals and plans will combine cross disciplinary modeling techniques from nonlinear dynamics, fluid mechanics, machine learning and bioinspired robotics.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.