If you’ve watched any James Bond movie with an underwater scene, you’ve likely seen 007 menaced by some form of the villains’ sinister undersea robots. In 2015, thanks to the efforts of Author, Biomimetics Researcher, and Neurophysiology Professor Joseph Ayers, undersea robots are a reality, and the future applications of his RoboLobster are far from evil.
Decoding the Lobster
Though the RoboLobster first came to life in the early 2000s, Ayer’s biomimetics research began as a graduate student in 1970, when he was studying the control of a lobster’s legs and their ability to use the same neural circuit to walk in different directions. After some additional research into the neural rewiring abilities of sea lampreys, Ayers discovered that both lobsters and lampreys have a command neuron, a coordinating neuron and central pattern generator (CPG) model of the organization of innate motor systems; he recognized that this model could serve as a potential blueprint for robots with comparable architecture.
“In the early 1990s, we built an undulatory mechanism for a lamprey robot, and then I was funded to build a lobster and a lamprey robot,” Ayers said. “When we first started building these, we used a control architecture called a ‘finite state machine’ and we organized the state machines on the rules of how these neural cogs operated. We built these robots and got them operating and quickly realized, as we used this algorithmic state machine architecture, unless we programmed an escape strategy for every possible contingency the robot could get into, it was gonna’ get stuck.”
Realizing the limits of existing autonomous technology, Ayers joined with an old post-doctoral advisor and some other physicists to develop nonlinear dynamical models of neurons. Discovering that lobsters have only four degrees of dynamical freedom, they came up with a set of equations that described interrelations of these degrees of freedom. From there came the realization that these electronic neurons could be used like “Lego parts” to model a nervous system. Add in a circuit design from his graduate school days, and Ayers had the early components to his robotic lobster.
“I realized we were on to something here. I think the most important point is, if you try to control robots with computer programs, unless you’ve anticipated everything they’re gonna’ do and have that wired into your controller, they’re gonna’ get stuck” he said. “Well, animals never get stuck…You can quickly convince yourself that their movements become chaotic and the beauty of these electronic neurons is they have ‘variable chaos.’ Our larger goal here is to start building these controllers with variable chaos, to see if we can get the robots to wiggle and squirm like the real animals.”
RoboLobster to the Rescue
While some people may look at the RoboLobster and see an expensive remote control toy, Ayers is quick to point out their practical applications. While he cites de-mining operations as a primary example, Ayers believes the potential of the robotic lobster and the robotic lamprey is almost as vast as the oceans where they’ll be put to work.
“What robots are good for is doing things that are too boring or too dangerous for humans to do. I think underwater mines certainly fits in the latter category,” he said. “A lot of mines are suspended in a water column and hung to an anchor then they float up at some depth. An undulatory robot (based on the lamprey) can swim through the water with a look-forward sonar and scan for these sorts of mines. That swimming robot, above a lobster robot (searching) down in clutter, such as rocks or coral, can also talk to the lobster and we can have a viable communications path.”
Despite the images presented in James Bond and other movies, the RoboLobster and lamprey are far from sinister. “People are very afraid of artificial intelligence because they see too many movies where people have great imaginations, but autonomy is the ability to go out and do something on your own. The autonomy we’re trying to create is called ‘supervised reactive autonomy,’” Ayers said.
That future of that autonomy, Ayers believes, includes robots that can literally sniff out explosives, re-charge their own batteries and even communicate with their operators anywhere in the world.
“We could send a ping out and determine when that robot arrived at a destination and triangulate where it is (at the bottom of the ocean). Then the central controller could send out a new search vector, which could be a compass heading and distance to walk on that heading,” Ayers said. “The robot walks out and does its thing pretty much autonomously, but when it gets to the end of its search vector, it would send another ping to a central station and that station would be connected via wi-fi to an operator onshore who could control this thing with an iPhone.”
Images: Jan Witting Photography