RoboJackets has returned to competition once again! Our RoboNav team took on the 28th annual Intelligent Ground Vehicle Competition (IGVC) this June at Oakland University in Rochester, Michigan.
RoboNav delivered an impressive performance: out of 9 teams competing in the AutoNav challenge, they placed 3rd in the design category and 4th overall. Jessiii completed more of the course this year than at the 2019 competition, and indeed more than any other RoboJackets entry to IGVC since 2012!
Now, you might be wondering what this course actually is. Traditionally, the IGVC obstacle course, being the primary challenge of the competition, is a track painted on grassy terrain with traffic barrels as obstacles. Teams aim to complete a full lap around the course without touching any obstacles or lines. In past years, only a few teams have been able to achieve this due to the visual processing and mechanical robustness required by the complex environment.
This year, the course was instead arranged on pavement to have it at the same location as the other parts of the IGVC competition. While this brought some advantages due to flatter terrain, it also tasked the team with preparing a new model for recognizing the lines with little time before the competition. The software team came together to collect and label data for lines on pavements in the SCC parking lot, and the newly retrained neural network was successful. Nonetheless, there were more preparations to be made even before going to the competition.
There were a number of hurdles that RoboNav had to persevere through just to start the event, let alone place 4th. Perhaps chief among these hurdles, as for so many other things, was the COVID-19 pandemic. For safety reasons, shop access has been limited in personnel and also in time — so, many team meetings only had 1 team member familiar with the software, and there could be no late-night debugging sessions at the shop in the weeks before competition. As a result, Jessiii had not run in over a year and still needed work.
Having the team together at the competition field gave them the opportunity for uninterrupted, collaborative, hands-on time to work. The team made good use of it by working throughout the day and gradually meeting the requirements to qualify and compete. The new wireless e-stop system, originally a new member project but untouched for longer than a year, was rebuilt and several issues were ironed out to get it working. They fixed the firmware controlling the motion of the wheels and the connections to the encoders, issues that had been plaguing the team since Jessiii’s last rewire. Batteries were repeatedly hauled to and from the pits in order to keep the debugging and testing going. By the final day of the AutoNav Challenge, these issues were worked out well enough to make Jessiii competitive. Kudos to everyone who made this happen!
While RoboJackets values working hard at competition, we also know that this doesn’t exclude having fun and making friends. For RoboNav at IGVC, the debugging sessions were reported to be a bonding experience within the team; only a few people could work on Jessiii at once, but the rest were having a good time and supplying support with other necessities like recharging batteries. RoboNav also made connections by walking around the pits and talking to other teams. After a friendly discussion about different robot design choices, one of these other teams even came to watch and support the RoboJackets design presentation.
The RJ Newsletter team asked Chuck Li, incoming RoboNav Project Manager, what the biggest team takeaways are from competition this year. It boils down to knowledge transfer, he told us. No one that went to IGVC this year had been to the competition previously, so it was hard to know how to prepare for design judging questions, how many tools to pack, or how to debug some of Jessiii’s systems in the field, for example. Thankfully, recent alumni were able to lend guidance in some cases, but RoboNav would like to implement a better process of passing down this kind of experience year over year.
In conclusion, RoboNav seems to have made the best of this competition. With a tried-and-true robot design but a lot of debugging and learning to do, the team came together to make it work.
If you’d like to see the design presentation that won the team 3rd place for yourself, you can find it below: