Space nerds love to gush about the mighty Curiosity, NASA’s one-ton rover currently rolling around Mars, drilling into rocks, and taking selfies. But last September, in a Massachusetts field, a long-necked robot built by students at West Virginia University in Morgantown did Curiosity one better: Working autonomously—that is, without direction from human controllers—the robot made decisions about where to search for various objects, located them (even ones it had never seen before), scooped them up with an ingenious shovel-and-basket device, and gently and precisely deposited them at a home base. Named Cataglyphis, after the desert ant whose navigation strategy it mimics, the robot was required to perform its tasks without using Earth-based navigation systems like GPS or magnetic compasses. (It used lidar and inertial systems.) Its success at the two-hour trial at Worcester Polytechnic Institute in Massachusetts won $750,000 for WVU and its Interactive Robotics Laboratory in the fifth year of NASA’s Sample Return Robot Challenge.
“It’s one of the most challenging things I’ve ever done, I’ll say that for sure,” says Morgantown native Nick Ohi, who was on the robot software team for all three years of WVU’s participation. “We were trying to show that we could do something that nobody else has been able to do yet.”
Team adviser Yu Gu, assistant professor of mechanical and aerospace engineering, points out that the Challenge was open-ended. “If a professor gives a problem to students on a test, the professor knows the problem can be solved,” he says. “But in the Challenge, there’s no guarantee the problem can be solved.” NASA has created 14 Centennial Challenges, including the robot challenge; a few expired after several years without a winner. The Challenges are open to anyone who wants to compete: small businesses, universities, individuals. “Challenges are NASA’s way of getting everybody to help with problems they do not yet have solutions to,” says Gu.
Using a fish-eye lens on the camera mounted on a mast, the robot could see 360 degrees to the horizon. The students wrote software to teach it to perceive its environment and to distinguish samples from sticks, rocks, and leaves. It learned how far away from a sample to attempt a pickup. If it failed, it would try again, at a different distance or from another angle.
After his team won the Challenge, Ohi used the experience to write a proposal for making planetary exploration robots smarter. The proposal won him a three-year National Science Foundation research fellowship, and he’s working on it there at the WVU Interactive Robotics Lab.
And after the big win, Cataglyphis has a bright future too. Gu is leading a research team to reconfigure the robot as a pollinator. It will detect and map bramble fruit—blackberry and raspberry—bushes. Instead of the scoop-and-basket, it will use an arm with soft bristles, mimicking the hairs on a bee, to deposit pollen in the flowers on the berry bushes.
There’s still time to enter NASA’s 3-D Printed Habitat Challenge, a competition to demonstrate a system for creating structural components from recycled or indigenous materials. The registration deadline is January 31, 2017.