Biologically derived approaches for the control and propulsion of swimming vehicles in riverine environments
The overarching objectives of this research program are to understand the sensorimotor and mechanical processes associated with maneuvering and active stabilization in swimming fish, and to engineer fieldable vehicle prototype that demonstrate swimming and maneuvering in riverine environments.
Bio-robotic fish (Anthony Mignano)
SAMUNO is a highly maneuverable bio-robotic fish developed to investigate underwater locomotion and control inspired by bluegill sunfish. The vehicle is capable of robust performance across a wide range of operating conditions, including both low- and high-speed swimming. SAMUNO can execute small-radius turns, rapid 180° course changes, zero-velocity turns, braking maneuvers, backing, station keeping, 360° rolls, and sustained swimming in flow.
These exceptional maneuvering capabilities are the result of knowledge developed over multiple research efforts, combining high-degree-of-freedom fin actuation with coordinated multi-fin control strategies. Our group has developed expertise in leveraging fluid–structure interactions, fin–fin coupling, body dynamics, and tuned stability to generate and control swimming forces in complex flow environments.
By integrating numerical modeling, machine learning–based control, and experimental validation, SAMUNO serves as a platform for studying how distributed fins and body motions can be coordinated to achieve maneuverability approaching that of real fish. This platform provides both a testbed for advancing bio-inspired underwater vehicle design and a foundation for future multi-agent and learning-based control studies.
Closed-loop velocity controller using reinforcement learning (Anthony Drago)
In this work, a high-fidelity numerical model of a bio-robotic fish was first developed to capture the system’s coupled body–fin dynamics. A reinforcement learning (RL) agent was then trained in simulation using this model and subsequently transferred to the physical robotic platform. This sim-to-real workflow enabled the development of a closed-loop velocity controller that coordinates fin and body motions, allowing the robot to accelerate to and maintain swimming speeds of up to 0.5 m/s in both simulation and experimental trials.