Robotics Learn From Fishes to Swim Better


Learning from nature to better solve human problems is an ancient practice. Biomimicry or biomimetics is the term for such practices and is defined as the imitation of the models, systems, and elements of nature for the purpose of solving complex human problems. This emerging field has brought us new technologies and ingenious solutions in various fields, and has seen stronger developments as we learn and appreciate the awe-inspiring nature, again and again, piece by piece. Recently, scientists in the robotics field have trained their robots to swim better, drawing inspiration from fishes.

Ghost knifefish is from rivers of the Amazon basin. They move around by undulating a long fin on their underside and feed in the night on smaller aquatic insect larvae and fishes. Black ghost knifefish and brown ghost knifefish are readily available as aquarium fish. The secret of its ability to navigate through the cluttered root masses and flooded forests in the darkness is a combination of three systems: active electrolocation enabled by its electric organ and electroreception organ, passive electrolcation used for communication purpose (a different electroreception organ is for this function) and a mechanosensory lateral line system to detect water disturbance caused by itself. Electric organ is derived from modified nerve or muscle tissue and is an organ common to all electric fish used to create an electric field. Highly sensitive sensory receptors are distributed over the entire body surface of ghost knifefish, able to detect a tiny change (one-tenth or one-hundredth of a millionth of a volt) in voltage of the electric field. Therefore, ghost knifefish can determine the presence of nearby objects by detecting electric field changes from all directions.

Currently, there are no underwater robots that can work well in debris-filled environment or in dark condition. These conditions are also very dangerous for human divers. Professor Malcolm Maclver in Northwestern University is building an underwater robot that can do rescue or survey work under such difficult conditions by learning both the locomotion system and the electrolocation sensory system from the ghost knifefish. One robot in Dr. Maclver’s lab can successfully move through obstacles in a tank, demonstrating artificial sensory capacity. A different robot, with a long fin at its bottom copying the special propulsion technique of the fish, can demonstrate locomotion capability. The next step is to bring these two capacities into one robot.

Dr. Maclver’s swimming robotics only learns the mechanism of how a fish navigate but a group of scientists went one step further to make their robotics look like fish, swim like fish and even feel like fish. Professor Andrew Marchese and his team in Massachusetts Institute of Technology published their paper this week in the first issue of Soft Robotics. This journal is dedicated to the study of soft robots, which are defined as not only having soft exteriors but also using fluid flowing through flexible channel as the power source.

The fish devised by Dr. Marchese’ team is the first ever self-contained autonomous soft robot able to execute rapid motions, including escaping maneuver, convulsing its body for direction change on a dime, and swimming quickly. The soft bodies of robots have multiple implications. First, with the increasing interactions between robots and human, the soft body is safe and presents no danger to hurt human if accidental collision happens. Second, instead of the traditional rigid kind of robots, for whom avoiding collision with the environment triumphs over efficiency of motion, the soft robot poses no danger to the environment or itself and can even learn to orient itself and reach to destination faster through making contacts with the environment. At last, the high maneuverability of real fish is only made possible by its continuous curvature of the body when it flexes, and this is impossible for rigid-body robot.

The movement of the tail of this robot fish is enabled by a long, tightly undulating channel on each side of its tail and the amount and duration of the carbon dioxide released from the canister in its abdomen. The current fish robot exhausts the carbon dioxide fairly quickly in straight line swimming, but it is fine since the purpose of building the robot was to study the performance capabilities, not for long term operation. A new version of the robot, expected to swim 30 minutes, will use pumped water instead of carbon dioxide for inflating the channels on the tail. One application of this longevity is to place it in schools of real fish to collect detailed information of their behavior in the natural environment.

The inspirations drawn from fish taught these two robots to swim better. Future studies from these two teams will likely produce even more fascinating news. Dr. Maclver’s underwater robot will learn to move and navigate like the ghost knifefish, providing invaluable information from dark and cluttered surrounding. The biomimetic soft robots ventures into a new frontier. Dr. Marchese’s fish shows that even with material whose response is uncertain, certain level of control is still possible. And this recognition is important for developing robots for the real, outside of factory, world.

By Tina Zhang


Science Daily
Nature Review

You must be logged in to post a comment Login