Robots learn to adjust its swimming speed to the current speed

Fish have a sensory system called lateral lines, which allows them to detect movement, vibration, and pressure gradient in the water. Scientists have now provided a robotic fish its own version of the system, so that it determines the best swimming speed.

The study involved researchers from Max Planck Institute for Smart Systems (Germany), Seoul National University and Harvard University. They created a mollusk-inspired robot that swims in place in front of the water flowing through the tank.

Its undulating swimming is accomplished by a series of linked silicone cavities, located along the sides of its body. Alternating air is pumped into the chambers on one side and out of the chambers on the other – this causes the inflated side to expand and curl outward, while the deflated side bends inward.

The robot’s lateral system consists of two silicone microchannel filled with liquid metal, running the lengths of each side. As each of those channels stretched while that side of the body curved, the resistance of the liquid metal inside increased. Therefore, by monitoring the change in drag, it is possible to determine the certain level of air pressure that caused the robot’s body to undulate.

The scientists set up a self-learning loop, in which a computer connected to a robot measures the velocity of water flow change, then automatically adjusts the air pressure in response to that information. . Doing so allows the robot to continuously maintain a swimming speed consistent with its current speed. In a natural environment such as a river, this prevents the robot from being swept downstream when not moving forward.

Dr Ardian Jusufi of Max Planck said: “This robot will allow us to test and refine the hypotheses related to the neuromechanics of swimming animals as well as help us improve our underwater robots in future”. “In addition to describing the characteristics of soft strain cells for the first time in submerged dynamic conditions, we have also developed a simple and flexible data-driven modeling method for controller design. our swimming feedback. “

An article about the research was recently published in the journal Advanced intelligent system. The robot can be seen in action in the video below.

Source: Max Planck Institute for Smart Systems