Making Waves: Upgrading Underwater Vehicles

Step into James Tangorra PhD’s Laboratory for Biological Systems Analysis at Drexel University, and you might think you’ve wandered into an aquarium. Sleek, mechanical sea creatures with articulated fins and flippers populate the space, their resemblance to living marine animals uncanny yet unmistakably robotic. This is the frontier of bio-inspired underwater vehicle design, where nature’s solutions are inspiring the next generation of submersible craft.

Nicholas Marcouiller, a PhD student,
Nicholas Marcouiller, a PhD student, works on SEAMOUR in Tangorrs’ lab.

The limitations of conventional underwater vehicles are clear to Tangorra, a professor of mechanical engineering and mechanics. “Most of the aquatic vehicles you see today aren’t too dissimilar from the quadcopters you’re familiar seeing in the air,” he explained. “They’re rigid and produce and control thrust to move slowly in tight spaces or they generate a lot of power to move quickly through calm water. None of them have the sensing that’s required to swim and maneuver within a surf zone.”

Current underwater vehicles perform crucial tasks like mapping the ocean floor, collecting environmental data, and exploring shipwrecks and geological formations. They’re essential for oceanographic research, often collaborating with other systems. However, their effectiveness is limited by poor performance in complex currents and diverse aquatic environments, hindering their ability to fully realize their potential in these important roles.

This is where bio-inspiration comes in. By studying the locomotion of marine animals, Tangorra and his team aim to overcome these limitations. “Animals endure multi-directional flow all the time,” he noted. “They know how to sense flow. They know how to work with currents. They know how to contour their body such that the forces that are being created by their body complements the forces that are being created by their fins or flippers.”

Marcouiller adds a foam block to SEAMOUR to adjust its buoyancy
during testing in Drexel's diving pool.
Marcouiller adds a foam block to SEAMOUR to adjust its buoyancy during testing in Drexel’s diving pool.

One of the stars of Tangorra’s research is SEAMOUR (Stroke Experimentation and Maneuver Optimizing Underwater Robot), a robotic sea lion that models the movements of its biological counterpart. SEAMOUR can roll, pitch and yaw by coordinating its flipper and body motions and dynamically adjusting its flipper gaits based on its swimming goals.

Recent findings with SEAMOUR have been promising. Using reinforcement learning, Tangorra’s team developed novel swimming gaits that outperformed the characteristic sea lion stroke in simulation and in real world swimming trials. This AI-driven approach allowed the system to explore a vast range of possible movements, optimizing for factors like velocity, heading stability, and maneuverability.

“The reinforcement learning-based approach not only showed success in generating an effective swimming gait in simulation but also retained its effectiveness when deployed in real world testing,” Tangorra explained. The learned gait demonstrated smooth, straight-line swimming when implemented on the physical system, consistently traveling farther than the characteristic stroke while adhering to a straighter path.

Especially interesting is how the computer-developed stroke was different from the sea lion’s natural movement. In a natural sea lion stroke, there are typically three main phases: recovery (where the flippers move out to the sides), power (where the flippers come together under the body), and paddle (where the flippers push water back). The AI-developed stroke essentially combined the recovery and paddle phases into one smooth motion, eliminating the distinct power phase.

Underwater Robotic Fin

The divergence from natural motion is a key advantage in combining bio-inspired design with mechanical methods. This approach allows the team to cherry-pick the best evolutionary adaptations while ignoring biological imperatives that don’t apply to robots. “We don’t have to mate. We don’t have to find food. There’s no sexual selection. We don’t have to fight against other predators,” Tangorra pointed out. “It’s a great example of how we can learn from nature, but also improve on it for our specific needs.”

Beyond SEAMOUR, Tangorra’s team has made significant progress in understanding fish swimming mechanics. Their research, published in the journal Bioinspiration & Biomimetics, revealed that the spacing between fins, their flexibility, and how they move in relation to each other all greatly affect swimming performance.

Using two different robotic fish and computer simulations, they discovered that even small changes in fin placement can dramatically alter swimming power. “With one of our robots, just moving the fins apart by about 50 mm changed how much forward thrust they created by up to 50%,” Tangorra explains.

The team also found that fin flexibility plays a crucial role. When they made the fins 30% more flexible, it not only reduced the overall swimming force but also changed when in the swimming motion the fins worked best.

Biological Systems Analysis robot fish drexel

“We learned that well-designed, interacting fins can produce several times more force than poorly arranged ones,” Tangorra notes. “It’s not just about copying fish exactly, but understanding why their fins work so well and applying that to our vehicles.” The implications of the team’s research extend beyond academic interest. As climate change continues to impact our oceans and coastal regions, the need to explore and better understand them becomes more and more urgent. As outlined in the National Oceanic and Atmospheric Administration’s FY22-26 Strategic Plan, there’s a push to “provide data, information and services to…accelerate growth of sustainable ocean industries and facilitate the technology advancements for coastal and marine solutions to climate challenges.” By learning from nature’s time-tested designs and optimizing them for specific tasks, Tangorra’s work may soon yield a new generation of underwater vehicles capable of navigating the complex and changing marine environments with unprecedented efficiency and adaptability.

This work is supported by key partners, including the Office of Naval Research and the National Science Foundation. These collaborations underscore the potential real-world applications of Tangorra’s research, from environmental monitoring to national defense.

As Tangorra and his team continue to push the boundaries of bio-inspired underwater vehicle design, the potential applications of their work grow increasingly exciting. From more efficient environmental monitoring to enhanced underwater exploration capabilities, these advances could revolutionize our understanding and stewardship of marine ecosystems.