CS Students Prevail in Planetary Rover-Landing Competition

5/26/2015 By Laura Schmitt, CS @ ILLINOIS

Three CS graduate students were part of a team that won a JPL-NASA-sponsored Space Design competition.

Written by By Laura Schmitt, CS @ ILLINOIS

In April, three CS graduate students were part of a team that won a JPL-NASA-sponsored Space Design competition to safely land a rover on Mars. Anna Jung-Chen Chen, Andrew Yi-Zong Ou, and Chi-Hsien Yen, along with ECE graduate student Jia-Bin Huang, made up the Guardians of the Galaxy team that beat out more than 15 other U of I teams to earn the $2,500 prize.

The Illinois team that won the Space Design Competition sponsored by JPL and NASA (from left): Andrew Yi-Zong Ou, Jia-Bin Huang, Anna Jung-Chen Chen, and  Chi-Hsien Yen. Huang is an ECE student. The rest are from CS @ ILLINOIS.
The Illinois team that won the Space Design Competition sponsored by JPL and NASA (from left): Andrew Yi-Zong Ou, Jia-Bin Huang, Anna Jung-Chen Chen, and Chi-Hsien Yen. Huang is an ECE student. The rest are from CS @ ILLINOIS.
The Illinois team that won the Space Design Competition sponsored by JPL and NASA (from left): Andrew Yi-Zong Ou, Jia-Bin Huang, Anna Jung-Chen Chen, and Chi-Hsien Yen. Huang is an ECE student. The rest are from CS @ ILLINOIS.

Given a photo of the Mars’ surface and a digital evaluation map image, the teams were challenged to find a landing site for a rover vehicle that was free of steep slopes and big rocks. According to Huang, while most teams devised machine-learning algorithms that learned to predict a safe landing site using training samples, he and his teammates devised a different solution.

“With our algorithm, we directly used physics and geometry to determine the safe landing zone,” said Huang, who brought computer vision expertise to the team. “With the other methods, if the algorithm went wrong, the teams wouldn’t know what happened. We also did a very comprehensive evaluation, indicating how many false positives we’d get in our algorithm.”

The Guardian team’s solution included an effective slope orientation estimation and hazard- and rock-detection algorithms that quickly calculated a solution in under 10 seconds and with 96% accuracy. In addition, the Guardians’ team created a full-frame confidence map that pinpointed the safest spot to land the rover. “The judges really appreciated this extra step,” said Ou.

While the team scored in the top third for their technical solution, they defeated the competition overall largely because of the professional caliber of their poster and presentation. “I really saw the passion from our team to win,” said Yen, who was primarily responsible for error analysis.

As a first-year graduate student working in CS Professor Lui Sha’s Integrated Emergency Cyber Physical Human Systems Lab, Ou learned a great deal from the experience. “The most valuable part of the experience was selecting a solution among many possibilities,” Ou said. “This is important for PhD students because we face options every day. We need to be able to select one direction with our research and believe that it is the right direction.”

Learn more about the team’s solution at the team's project page.


Share this story

This story was published May 26, 2015.