Mobots compete at Carnival
Booth. Buggy. Carnegie Mellon has many unique Carnival events, and one of the less prominent — but no less unique — events is the annual Mobot races held by the School of Computer Science (SCS). Robotics Institute senior research technician Greg Armstrong emceed the event in true Scottish fashion, wearing a kilt with the Armstrong clan tartan and carrying a sword for effect.
The 17th Annual Mobot Races took place last Friday afternoon on the course outside Doherty and Wean halls. The competition challenges teams to build and race their vehicles (“MObile roBOTs”) along a solid white line through a series of gates arranged in a slalom-like pattern (see last week’s SciTech for more details). In addition to SCS, the Mobot events are sponsored by Lockheed Martin, TwoSigma, Google, and Misumi USA. The winner of this year’s competition and the $1,000 grand prize was Team Stingray, a one-man team consisting of junior Kwabena Agyeman, an ECE major. The second-place prize of $500 went to Team Commandaria, which consisted of senior math and computer science double major Aaron Jaech and senior ECE majors Diana Hu and Kyle Neblett. Third place ($250) went to Team Three Dollars to Win, comprised of sophomore ECE major James Wahawisan.
Despite winning first place out of the six undergraduate teams competing, Stingray did not complete the Mobot course. The course has a total of 14 gates, and Stingray maneuvered through eight, making it the furthest in the shortest amount of time, before running off the line into the grass. Each mobot is programmed by its designers, and Agyeman did not include an algorithm to navigate the challenging second half of the course, in which the robots must choose to take the left or right fork at several “decision points.”
These decision points are only a few of the many challenges on the Mobot course. While humans have the option of taking stairs on the sidewalk down the Mall, the mobots must drive down the two steep hills to finish the course. Robots that go down a hill too quickly will lose track of the path or crash, making designers take this into account when building and programming their mobots. For instance, Agyeman said that his “robot’s pretty much driving backwards” when going downhill.
In Friday’s race, the bright sunlight created difficulty for the mobots’ vision sensors. During the awards ceremony after the races, Catherine Copetas, the SCS assistant dean for industry relations, SCS director of special events, and a member of the Mobot organizing committee, commented that “sensors just don’t like really, really bright light.”
The resurfaced patches of cement on the sidewalk also challenged the mobots. Although most of the sidewalk is a dark gray color, certain patches have been replaced with a lighter color. Because these patches are close to the white color of the track, many mobots drove off the course after reaching them. A patch just after the second gate proved particularly difficult.
In addition to the six undergraduate teams competing for the awards, three exhibition teams also ran the course. Eli Richter’s mobot, named Johnny 0.5, was the only one to finish the entire course Friday, completing all 14 gates in a time of 1:00.30. Richter said that his mobot has sensors with resolution as high as 1/10-inch and that his main limitation on time was that he “just wasn’t able to steer as tight” given the physical constraints of his robot’s frame.
While the focus of Friday’s event was on the competition, the Robotics Club members used the Mobot races to raise awareness of their group. Several competitors were in the club, and the group was selling T-shirts nearby. Red Robot, the Robotics Club’s mascot, was wandering through the crowd to the delight of the many children at the event.
The Mobot Committee has plans to expand the competition beyond the existing path-following challenge. Copetas said that they plan to add the challenge of following the course backward after it is completed and that she would like to incorporate the Gates Center’s Helix into a Mobot competition in the future. She stressed that mobots’ small size or the seemingly simplicity of their task is deceptive. “We should not underrepresent how much work goes into mobots,” she said.