The Fascination of Technology Robot Fish Adjusts Its Size

Related Vendors

Ecole Polytechnique Fédérale de Lausanne (EPFL)

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

SIMTOS (Seoul International Machine Tool Show)

In our “Fascination with Technology” section, we showcase impressive research and development projects for design engineers every week. Today: A robotic fish that can explore everything from shallow streams to open waters.

Propeller-driven underwater vehicles have long assisted scientists in researching and monitoring marine environments. However, their mechanical design imposes limitations: Rotating propellers can get caught in vegetation, stir up sediment, and startle the animals they are actually meant to study. As a result, they are only partially suitable for shallow streams, densely vegetated bodies of water, or observing fish at close range.This is one of the reasons why robotics experts have been developing machines for years that move like fish—by bending their bodies instead of rotating a propeller. The challenge is that most fish-inspired robots are designed for a specific size and a specific task. Scaling them up or down therefore usually means starting from scratch.

If you want to monitor a stream and then a lake today, you basically need two different robots, each of which must be developed and tested from scratch.

Nana Obayashi

A team of engineers from EPFL and New York University (NYU) believes it has found a solution to this problem. The researchers introduced ScaFi (short for “Scalable Fish”), a robot modeled after fish such as cod and mackerel. These fish concentrate the greatest curvature of their bodies in the area near the tail. In nature, this mode of locomotion works across an unusually wide range of body sizes.“If you want to monitor a stream and then a lake today, you basically need two different robots, each of which has to be developed and tested from scratch,” says Nana Obayashi, first author of the study published in npj Robotics. “The environments we’re interested in don’t come in one size—so neither should our tools.”
Obayashi is currently an assistant professor of mechanical and aerospace engineering at the NYU Center for Robotics and Embodied Intelligence. She led the project during her Ph.D. at the Computational Robot Design & Fabrication Lab, headed by Josie Hughes, in the School of Engineering at EPFL.

ScaFi has a rigid front section and a flexible tail made of fiberglass rods. A single motor pulls on two cables that cross near the tail end. This creates the S-shaped curvature required for the fish-like swimming motion. 
The diameters of the rods that make up the tail are the only components that need to be adjusted to the robot’s size. As the robot is scaled up, the rods become proportionally thicker to ensure comparable bending behavior of the tail. The underlying drive mechanism and the system of crossed cable pulls remain unchanged. This is significant because it could reduce the development effort required for fish-like robots designed for different environments. At the same time, researchers gain a platform with which they can systematically investigate how swimming performance changes with size—a question that is difficult to study systematically in both animals and custom-built robots.

Let's Hit the Water!

The team built three robots measuring approximately 0.6 (1.97 feet), 1.1 (3.61 feet), and 2.9 meters (9.51 feet) in length and tested how well they could swim. In collaboration with the EPFL Unsteady Flow Diagnostics Lab, led by Karen Mulleners, the researchers found that the smallest robot generated vortex patterns in the water similar to those of real fish. For all three sizes, the swimming motion closely matched that of real fish after adjusting for body size. This suggests that the scaling approach preserved the fish-like movement pattern even though the robot’s length increased nearly fivefold.
The researchers also tested the robots under real-world conditions: the medium-sized robot in a Swiss stream, the largest in Lake Geneva, and the smallest in streams only 15 to 30 centimeters (5.9 to 11.8 inches) deep. During the test in the stream, the robot continued to swim even after the GPS connection failed. However, energy efficiency proved more difficult to scale. The two smaller robots achieved similar results, while the largest was consistently less efficient and required a different, more powerful motor. The authors suspect that flow resistance and inertia may be responsible for this. The exact cause, however, remains unclear. The team was thus able to scale the swimming motion itself more reliably than the energy expenditure required to generate it.
A similar trade-off emerged in disturbance tests. The smallest robot was the most maneuverable but took the longest to stabilize after a course deviation. The larger robots were less maneuverable but more stable. The same approach, in which scaling is based on a central structural parameter, could also be applied to other compliant robots—including those used outside of water. Whether energy efficiency can be scaled as successfully as the swimming motion remains an open question.

Fascination with Technology

Inspired by the Puffin: Robot Flies, Swims, and Dives