WVU team preps for upcoming Sample Return Robot Challenge

From left: Jared Strader, Nicholas Ohi and Matthew Gramlich.

At the 2015 Sample Return Robot Challenge, the team of engineering students from West Virginia University and their robot, Cataglyphis, came incredibly close to snagging the $1.5 million prize offered as part of NASA’s Centennial Challenge’s.

While the team did make history last year by being the first to successfully retrieve the coveted “red rock” sample, earning them $100,000, the Mountaineers have their sights on one thing for 2016: finishing the job.

They intend to try and do just that as they travel to the campus of Worcester Polytechnic Institute in Massachusetts, September 2-5, for the final year of the event, where they will face stiff competition with six teams standing in their way.

The biggest uncertainty, according to team advisor Yu Gu, assistant professor of mechanical and aerospace engineering , is time and sunlight.

“They need to fit seven teams into two days, with each run taking two hours on the competition field,” Gu said. “That means that sample searches may have to happen in the early morning or late evening, when lighting conditions are challenging for computer vision algorithms.”

Unlike previous years, the teams have been given advance notice as to the approximate regions where samples will be placed. WVU has put this to good use as it sought to make significant changes to Cataglyphis, which is named after a desert ant known for its ability to journey across great distances and reliably return home. One of the biggest improvements, according to software co-lead Nick Ohi, can be found in how the robot plans its actions.

Last year the robot’s decision-making was mostly scripted beforehand and it could only make small decisions to deviate from its main sequence. This year, Ohi said, it is up to the robot to decide what it should do.

“Last year, the robot was only able to search location one, then location two, then location three and so on in a very predefined order and it would waste a lot of time searching areas that contained no samples,” Ohi said. “Now, it’s smart enough to make these decisions on its own and improve how it spends its time. The field is very large and the robot has a lot of ground to cover in only two hours, so spending its time efficiently is key.”

The robot now has the ability to choose what it believes to be the most effective way to spend its time searching for samples.

“The robot considers all of the possible behaviors it can perform and chooses to execute the one it believes gives it the best chance of succeeding in its mission,” said Ohi, a Morgantown native. “It makes these decisions based on data it has collected about possible sample locations on the field, places it has already searched and knows it does not need to go back to and the difficulty of the terrain it is trying to traverse. All of this data gets stored in a big map of the field onboard the robot’s main computer. We have given the robot an initial set of data before it goes out on the field, but as it drives the field and searches for samples, it collects additional data with its camera and 3D LIDAR sensors and it’s able to improve its decision making with this new data in real time.”

“From our perspective, I think the redundancies in our navigation system and mission planning will be a boon since the robot has plenty of opportunities to update its internal map based on what its sensory systems are returning,” said Lisa Kogan, mechanical team lead from Toronto, Canada.

The objective of the challenge is to encourage innovations in autonomous navigation and robotic manipulation technologies. These innovations may enhance NASA's space exploration capabilities and could have applications on Earth, continuing the nation's leadership in robotic technology.

NASA Centennial Challenges were initiated in 2005 to engage the public in the process of advanced technology development. The program offers incentive prizes to generate revolutionary solutions to problems of interest to NASA and the nation. Competitors are not supported by government funding and awards are only made to successful teams when the challenges are met.

Joining Ohi and Kogan on the team include returning members Kyle Lassak of Loretto, Pennsylvania; Matt Gramlich of Hurricane; Scott Harper of Spencer; Jared Strader of Clarksburg; Boyi Hu of Qinhuangdao City, China; and Alexander Hypes of Lewisburg. New members include Nathan Tehrani of Morgantown; and Chizhao Yang of Jiangxi, China. Joining Gu as faculty advisors are Powsiri Klinkhachorn of the Lane Department of Computer Science and Electrical Engineering and Jason Gross and Marvin Cheng from the Department of Mechanical and Aerospace Engineering.

The team is sponsored by the Statler College of Engineering and Mineral Resources , the Department of Mechanical and Aerospace Engineering, the Lane Department of Computer Science and Electrical Engineering, NASA West Virginia Space Grant Consortium and the West Virginia High Tech Consortium Foundation .

---