The Resizable Robotic Fish Built for Any Body of Water

Robotic fish are useful because they can go places that conventional underwater vehicles cannot. They can swim through vegetation without worrying about snagging a propeller, move through shallow water, and get close to wildlife without causing much of a disturbance. But a robotic fish designed for a narrow creek usually isn't much use in a larger body of water, like a lake. That means different robots need to be designed for each environment.

Now, researchers have developed a robotic fish called ScaFi that makes moving between environments far easier. Instead of being designed around one particular size, ScaFi can be scaled from around half a meter to nearly three meters long while keeping essentially the same mechanical design. That makes it possible to build smaller versions for confined waterways and much larger ones for open water without starting over each time.

ScaFi's swimming is powered by a relatively simple mechanism. A rigid front section houses the electronics and drivetrain, while the rear consists of a flexible tail built around fiberglass rods. A single motor operates two tendons routed through the tail in a crossed configuration. As the motor moves them back and forth, they bend the tail into an alternating S shape that produces the undulating motion needed to propel the robot through the water.

The key to making that mechanism work at different sizes came down largely to controlling the stiffness of the tail. The researchers developed a scaling model that determines how thick the fiberglass rods should be based on the robot's length. Larger versions receive thicker rods, while smaller versions use thinner ones. This allows the tail to maintain roughly the same bending behavior as the overall robot changes size, without adding more motors or substantially changing the basic mechanism.

The team tested the idea by building 0.6-, 1.1-, and 2.9-meter-long versions of ScaFi. Particle Image Velocimetry experiments with the smallest version showed that its tail produced rotating vortices behind the robot similar to those associated with biological swimming. Tests across all three sizes also showed that the scaling approach was able to preserve similar tail kinematics despite the nearly fivefold difference in length.

The variety of designs is important, because each size has advantages in different environments. The smallest ScaFi was deployed in creeks just 15 to 30 centimeters deep, where a larger robot would risk hitting the bottom. The medium version was tested in a flowing stream, and the 2.9-meter model was deployed in the open water of Lake Geneva, where its larger size provided greater stability and payload capacity.

However, scaling the robot in this way does come with some compromises. The largest ScaFi was considerably less energy efficient and required a different motor, with increasing mass, inertia, and hydrodynamic drag creating new challenges. Larger versions were also more stable when disturbed, while the smaller robot was more agile but took longer to recover after being knocked off course.

Despite these differences, ScaFi provides a common starting point for building swimming robots around the environment in which they need to operate. Instead of developing a new machine every time the scale changes, engineers can resize the same basic architecture and have a much better idea of how it will behave once it hits the water.