ScaFi scales a research fish robot from 2 feet to nearly 10

At the water’s edge, ScaFi’s red fabric tail lies behind a rigid, motor-driven front section. The team from NYU Tandon built the fish-inspired robot at three lengths—from 0.6 meters (2 feet) to 2.9 meters (9.5 feet)—using one underlying design, then put the machines into creeks, a Swiss stream and Lake Geneva.

Propellers are awkward in the places scientists often want to observe: their blades can catch vegetation, stir sediment and startle wildlife. ScaFi bends instead. One motor pulls two tendons that cross near the end of a flexible fiberglass-rod tail, creating the S-shaped motion used by fish such as cod and mackerel. As the robot grows, only the rods’ diameter changes; the motor mechanism and crossed-tendon system stay the same.

That recipe held up across the size range. After adjustment for body size, the three robots’ swimming motions lined up closely, even as their length increased nearly fivefold. The smallest produced swirling water patterns similar to those left by real fish. In the field, the medium robot swam in a Swiss stream, the largest in Lake Geneva and the smallest in creeks just 15–30 centimeters (6–12 inches) deep. During one stream test, it continued swimming after losing GPS — the satellite-positioning signal.

The limitation is energy. The two smaller robots performed similarly, but the largest was consistently less efficient and needed a different, more powerful motor. The researchers suggest drag and inertia may be responsible, although they have not resolved the cause. Size also changes handling: the smallest robot was most agile but recovered more slowly after being knocked off course, while the larger machines were less nimble and more stable.

So what changes in practice? A team monitoring a creek and then a lake could start from the same tail architecture rather than designing and testing two robots from scratch. That makes ScaFi a useful research platform for aquatic monitoring and for studying how swimming changes with scale. It is a research platform: the tests show that motion scales cleanly, not that energy use, autonomy or deployment costs do. The findings were reported by the team in npj Robotics.