Shirly Manor’s Post

I’m excited to announce that we won 1𝐬𝐭 𝐏𝐥𝐚𝐜𝐞 at the 𝐀𝐮𝐭𝐨𝐧𝐨𝐦𝐨𝐮𝐬 𝐊𝐚𝐫𝐭𝐢𝐧𝐠 𝐒𝐞𝐫𝐢𝐞𝐬 2026 𝐔𝐒 𝐍𝐚𝐭𝐢𝐨𝐧𝐚𝐥 𝐆𝐫𝐚𝐧𝐝 𝐏𝐫𝐢𝐱. Last week, the Purdue Electric Vehicle Club’s Autonomous Project Team won 1𝐬𝐭 𝐨𝐯𝐞𝐫𝐚𝐥𝐥 against 9 teams from across North America, including: 🏁1st in the qualifying rounds 🏆1st overall ⏱ 1:03 fastest lap (6:20 official 5-lap time) As a 𝐫𝐨𝐨𝐤𝐢𝐞 𝐮𝐧𝐝𝐞𝐫𝐠𝐫𝐚𝐝𝐮𝐚𝐭𝐞 𝐭𝐞𝐚𝐦, we built this kart over the span of a year, designing and integrating the full mechanical, electrical, and software stack. Every member had to focus on many parts, switching from fixing our control algorithm to working on the custom PCB and designing and integrating a chain-drive steering setup. Our unique advantage came from integrating multiple sensor inputs at different frequencies into a single localization system, paired with dynamic, tuned control algorithm(s) operating in parallel for fast updates and a learned system. We experimented with 𝐌𝐏𝐂 and training residual models (small 𝐑𝐍𝐍𝐬 always loaded into VRAM at inference with dynamic training + checkpointing on new data) to achieve smoother control and to learn the sim-to-real gap. Integrating this with the gray-box world model for more accurate simulation from collected data allowed for better testing and integration while off-track, running millions of laps before ever testing on the kart. We also presented at the 𝗡𝗲𝘅𝘁-𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗼𝗻 𝗧𝗿𝗮𝗻𝘀𝗽𝗼𝗿𝘁 𝗦𝘆𝘀𝘁𝗲𝗺𝘀 𝗖𝗼𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲, where we received technical feedback from professors and graduate students at many universities and collaborated with other teams to design systems for next year (aiming for 𝐦𝐮𝐥𝐭𝐢-𝐤𝐚𝐫𝐭 racing). None of this would ever happen without our 𝐚𝐦𝐚𝐳𝐢𝐧𝐠 technical team. Thank you to Shriya Balu, Duel M., Daniel Proano, and, of course, my co-lead, Millan Kumar. Thank you to everyone in EVC for making sure this event was successful, including but not limited to Jack Thornton, Nathan Pope, Rachel White, Shae Stonemeier, and Philip Goldberg. We could never have done it without you. Thank you Andrew G. and the Autonomous Karting Series Inc. team for all of the work putting on this event. It was truly special to get the whole community together and I look forward to racing next year! As a rookie team, this is our first iteration, and we are committed to competing harder and faster next year. 𝐄𝐕𝐂-𝐀27 is targeting a stronger MPC with faster inference, 𝐬𝐮𝐛-50-𝐬𝐞𝐜𝐨𝐧𝐝 lap times, and wheel-to-wheel racing against a drivered kart, in an attempt to run a full multi-kart autonomous race next year. Check out our 𝐄𝐕𝐂-𝐀26 code here: https://lnkd.in/g2SYS_pf We’re recruiting strong technical Purdue students for next year. Reach out if you’re interested!