One Blueprint, Fivefold Growth, a Length-Scalable Fish Robot Built for Creeks and Lakes Alike

Most fish-inspired robots arrive locked to one length and one job. Scale them up for a lake or down for a creek and the whole machine usually has to be designed again from scratch. Engineers working between NYU Tandon and EPFL have shown a way around that limit with ScaFi, a compliant robot modeled on the tail-heavy swimming of fish such as cod and mackerel. Those species keep most of their body bend toward the rear, a style that works across a wide range of real animal sizes. ScaFi copies the same idea with a rigid motor-driven front and a flexible tail of fiberglass rods wrapped in red fabric. A single motor pulls two tendons that cross near the tail tip, forcing the familiar S-curve that pushes water backward.

Only the rod diameter needs to change as the robot grows. A scaling rule based on beam theory and hydrodynamics determines how thick the rods become. This ensures that the tail bends around the same amount regardless of whether the robot is small, medium, or huge. The engine, tendon layout, and overall design remain the same. They developed three prototypes of the robot: 0.6 meters long / 1.5 kilos, 1.1 meters long / 2.4 kilograms, and 2.9 meters long / 11 kilograms, representing a fivefold increase in length. Once scientists adjusted the swimming motion for the body size, the gait was quite consistent among all three. Particle-image testing on the smallest robot revealed that it produced swirling wakes that resembled the reverse von Kármán streets created by real fish.

Out in the field, each size of robot was deployed where it belonged. The smallest one could swim across creeks as shallow as 15 to 30 centimeters and past weeds and banks without getting snagged like a spinning propeller. Even when the GPS signal failed, the medium robot remained on course in a Swiss stream. Even in the face of currents and waves, the largest robot performed admirably on Lake Geneva. Nana Obayashi, who worked on the project as a doctorate researcher at EPFL and is now an assistant professor at NYU Tandon, put it simply: if you want to watch a creek and then a lake, you need two different robots, each created from scratch. Environments don’t come in one size, so why should the tools used to study them?

Energy consumption did not scale as well as the movement itself. The two smaller robots required almost the same amount of power to complete their tasks. The 2.9-meter version required a more powerful engine and still consumed more energy, most likely because drag and inertia increased more faster than the scientists anticipated. If you knocked the smallest robot over, it was slow to recover while being the most agile; the larger ones were steadier but less nimble. These trade-offs correspond to what biologists find in real fish of varying lengths. So researchers now have a single platform, ScaFi, that can be created to the exact size of the water they’re researching, eliminating the need to start over every time the environment changes. The similar rod-thickness criteria they adopted could be useful for other flexible robots that must operate in a variety of sizes.
[Source]