Sea turtles are amazing creatures, and if you’ve ever seen one in the wild, you’re very lucky. If you never get a chance to see a sea turtle up close, though, you still might be able to get a good look at a C-Turtle. Unlike their shelled counterparts, C-Turtles are desert creatures, though they can adapt to many environments from Mars to a living room. These flat robots are the creation of a team of researchers from Arizona State University, and while they might not swim, they can crawl through all sorts of challenging materials.
The idea for C-Turtle came about because the researchers wanted to create inexpensive, easy-to-make robots that could interact with real-world environmental conditions. Their work was documented in a paper entitled “From the Lab to the Desert: Fast Prototyping and Learning of Robot Locomotion,” which you can read here.
According to the researchers, simulated conditions to “train” robots in labs often don’t equate to conditions in the real world, rendering that training less than optimally effective. In addition, those simulations are often used to iterate robots, meaning that their final designs and behavior end up being based around their performance in simulated rather than real world conditions. Therefore, they decided to develop a robot that could be manufactured in less than a day’s time, tested immediately in the real world, and allowed to adapt to those real world conditions.
“In this work we argue that the design of effective locomotion strategies is dependent on the interplay between a) the shape of the robot, b) the behavioral and adaptive capabilities of the robot, and c) the characteristics of the environment. In particular, adverse and dynamic terrains require a design process in which both form and function of a robot can be rapidly adapted to numerous environmental constraints,” the researchers explain.
They designed a robot whose motion was modeled on that of a sea turtle’s crawl, and used laminate manufacturing to create the body of the C-Turtle. Laminate manufacturing, as defined by the researchers, is a process that involves “planar sheets of material that are iteratively cut, aligned, stacked, and laminated to form a composite material.” The five layers of the robot were laser-cut and then fused together using a heated press. Motorized, interchangeable fins were created, and the motors were attached to the paper body by means of 3D printed servo horns, which, according to the team, took as long to print as the rest of the C-Turtle took to create.
Still, the entire C-Turtle took only an hour to make, meaning that the team could quickly assemble and test new iterations. Machine learning tests were run on the robots both in a natural outdoor desert environment and in the lab, using poppy seeds to simulate sand. Early iterations demonstrated problems such as a tendency to burrow into the sand, so changes were quickly made to the design. Four different fin designs were tested: two modeled on natural sea turtle fins, and two based on synthetic shapes.
“The results…indicate that for every fin that underwent learning, in both artificial and natural environments, the final locomotion policy shows some degree of improvement with regard to distance traveled by the robot after only 10 iterations,” the researchers state.
Overall, the biologically inspired fins out-performed the artificially inspired fins. The researchers also noted that the robots performed differently in the natural sand environment and in the poppy seeds, because of slight differences in the density and size of the granules. The natural environment itself fluctuated over time, however, due to inconsistencies of moisture, but an algorithm programmed into the robots allowed them to adapt quickly to changing surfaces.
The study revealed that the team’s laminating and 3D printing method was well-suited to the fast production of robots that could be constructed and taught new behaviors in less than a day. That’s extremely fast compared to how long it typically takes to build and train a robot, and the approach allowed the researchers to have their machines out in the real world much sooner than they otherwise would. Because of their adaptability – and because they’re so quick and cheap to make – the team believes that C-Turtle robots could be effective in sweeping deserts for land mines.
Authors of the study include Kevin Sebastian Luck, Joseph Campbell, Michael Andrew Jansen, Daniel M. Aukes and Heni Ben Amor. You can learn more about C-Turtle here. Discuss in the C-Turtle forum at 3DPB.com.