Something’s Fishy About This Robot

We know our own control system better than we know any other, so it is no wonder why roboticists tend to roughly mimic the human nervous system when designing their machines. A computer might take the place of the brain, and wires and sensors take the place of the spinal cord and nerves, but the same basic architecture, with a central processing unit, is apparent in both cases. It may seem obvious to us that this is the best way to go, but perhaps it would be worth taking a step back to reevaluate that assumption.

Consider the eel, for example. These sea creatures have a more distributed neural architecture that enables them to keep on swimming even if their spinal cords are completely severed. Not only does their unusual method of locomotion allow them to survive serious injuries, but it also makes them highly adaptable in a variety of environments. Both of these traits would be very useful in robots, but designing machines with these capabilities is extremely challenging.

So rather than starting from scratch, a group led by researchers at Tohoku University decided to borrow some ideas from the biology of eels. They have designed and built an eel-like robot that has multiple types of sensing units distributed around its body. These sensors do not need to send their data to a central control unit for processing, but rather can directly interact with their environment to produce coordinated locomotion.

The researchers integrated two kinds of sensory feedback — stretch signals from the body and pressure feedback from the skin — into a mathematical model of neural circuits. In simulations and robotic tests, these distributed loops quickly produced stable swimming patterns without heavy reliance on centralized processing.

The model assumes that each body segment has its own local control unit, much like the biological central pattern generators found in vertebrate spinal cords. These local circuits are capable of producing rhythmic movements on their own but are further stabilized and coordinated by incoming sensory feedback.

A robotic eel built by the team using their control system not only swam effectively in water, but also managed to crawl across land and maneuver around obstacles. Stretch feedback proved particularly important on land, as it enabled the body to push against fixed objects to generate forward thrust. This is important because it shows that the same underlying control scheme that drives aquatic swimming can also be repurposed for terrestrial locomotion.

This work proves that by embracing distributed architectures and local feedback loops, roboticists can create machines that are more robust, adaptable, and capable of thriving in uncertain environments. Eels may not look like much of an engineering role model at first glance, but as this study shows, their neural design could help us to create the next generation of robots.