With the development of technology, soft robots are no longer a new species. So, what kind of scenario would it be if a robotic fish could automatically adjust to the speed of the water while swimming in the water?
Recently, a research team composed of Germany's Max Planck Institute for Intelligent Systems (MPI-IS), Seoul National University in South Korea and Harvard University in the United States successfully developed an underwater soft robot. The robot can not only swim underwater like a fish, but also automatically adjust the swing amplitude in the water according to the speed of the water.
The team’s paper was published in the journal AdvancedIntelligentSystems under the title of "Modeling and ControlofaSoftRoboticFish withIntegratedSoftSensing" ("Modeling and ControlofaSoftRoboticFish withIntegratedSoftSensing").
To design the controller, the research team developed a data-driven, lumped parameter method. This method allows accurate but lightweight simulations using experimental data and genetic algorithms. The model can accurately predict the behavior of the robotic fish at driving frequency and pressure amplitude, including the effect of antagonistic co-contraction on soft actuators.
The two sides of the robotic fish are composed of artificial muscles made of silicone chambers. The researchers deliver air on both sides of the robotic fish. When one side expands, it will bend outwards, and the air pockets on the other side will contract and bend inwards. This will encourage its tail to swing to the right and left, which is the same as the movement pattern of a real fish swimming in the water.
In order to improve the sensitivity of the sensor, the researchers pre-stretched the soft strain sensor to the length of the robot, and used tape and cable ties to firmly connect the two ends of the sensor to the robot. The researchers embedded the most advanced software strain sensor to measure the bending of the robotic fish. The silicone microchannel of the sensor is filled with liquid metal, which can be as flexible and stretchable as a telephone line.
When one side of the robot fish's body is bent, it indicates that the resistance of the liquid metal is increased. In order to measure resistance, the researchers connected a sensor to an ohmmeter, and determined the magnitude of the robot fish's body fluctuations due to a given amount of air pressure by monitoring the changes in the resistance data.
The researchers put the robotic fish in the water tank to test the air pressure controller, and confirm the swimming performance of the robot through the data of the sensor. They found a signal loop that provides a self-learning algorithm to the controller.
Through this algorithm, when the right amount of air passes through the pneumatic device, it can be automatically adjusted. In this way, the robotic fish can maintain a swimming speed adapted to it according to the water flow speed. This means that the robotic fish will not be washed away even if it does not move forward in environments such as rivers.
"We have developed a fluid dynamics model that can predict the behavior of robotic fish. This is based on previous research. We measured the peak thrust of the robotic fish when swimming in a contracted and stiff state, and tested the feed forward undulating motion. The soft sensor. Proprioceptive soft sensory feedback enables the robot to respond to different flow conditions.” explained Ardian Jusufi, the motion project leader of MPIIS's biorobot and somatic cell system and the corresponding author of the paper.
"In this work, we adopted a simple method to build a data-driven model of the soft robotic fish and extend it through controller design. The model can be easily extended without the need to completely rebuild it. For example, research The zoom effect of the robot or test different types of sensing technology." Yu-Hsiang Lin of MPI-IS, the first author of the paper, described the motion in biological robots and somatic cell systems.
Researchers believe that the sensor is a brand-new design technology and described it as a "superelastic liquid metal strain sensor" in the paper. The technology was jointly developed by the team of Professor Yong-Lae Park of Seoul National University and Daniel Vogt of Harvard University.
"Through biology and soft robotics technology, we have a soft robot with unlimited degrees of freedom. This means that any part of the soft robot body can be deformed. It is difficult for us to predict how the shape of the fish will change because we cannot be sure. The shape of a fish. There are many sensors installed on the human body because only rigid robots with a limited number of joints can do this."
In an interview with foreign media, Jusufi said: "This robot will allow us to test and refine hypotheses about the neuromechanics of swimming animals, and help us improve future underwater robots. In addition to the first characterization of soft strain sensors under underwater dynamic conditions. , We have also developed a simple and flexible data-driven modeling method to design our swimming feedback controller. This model will arouse future work in the scientific community and will help speed up the design and operation of soft robots. In our research, we will also use soft strain sensors in land-based robots, which can climb and overcome complex obstacles."
Constructing robots by studying the swimming laws of fish is conducive to the design of artificial soft structures. In the future, this robot may continue to explore the depths of the ocean and provide more valuable marine biological data. It not only avoids noise, but also reduces the risk of entanglement faced by traditional propulsion submarines.
Robot fish may become a new energy-saving choice, which is why more and more researchers are investing a lot of energy in the development of soft actuators and sensors. In a nutshell, this paper is an important research result in the field of fish neuromechanics and morphological intelligence, and it may accelerate the development of the biological field.
