Congratulations to TUM for winning the $1M Indy Autonomous Challenge … with Galactic!! The winning team recorded the fastest 2-lap average speed of 218.78 km/h, 60.77 m/s on the famed IMS Oval. It was an exciting and emotional afternoon for the teams and huge crowd with 350 high school STEM students (who all got the coveted IAC racing turtle t-shirts) and thousands of spectators in person watching the university teams do speed trials and obstacle avoidance contests. Our friend Mahendra Bairagi of AWS Sports arranged the Twitch IAC livestream watched by over 36,000 people including our ROS community.
“Open source collaboration got us here” said Paul Mitchell IAC leader and president / CEO Energy Systems Network twice during the awards ceremony. Paul then announced that Alex and the TUM team are open sourcing all of their work. Here is the TUM team’s GitHub with the winning IAC racecar’s:
- Graph-based local trajectory planner for race vehicles in a dynamic environemnt
- Scenario Architect for the creation of short scene snapshots for autonomous driving benchmarks
- Optimization algorithm for the creation of global, optimal raceline
- Path and velocity controller for an autonomous racecar
- Neural Network for Object Detection with Camera and Radar Data
- Vehicle dynamics simulation for autonomous vehicles
- Library with functions for trajectory planning
- Database with maps of international racetracks
- Quasi-static laptime simulation
- Recursive Uncertainty Model Identification
Open Robotics, Autoware Foundation, Eclipse Foundation and our ROS community supported all the teams via the IAC Base Vehicle Software WG with Alexander representing all university teams and co-led by Gina O’Connell of IAC, Josh Whitley, Neil Puthuff and Joe Speed.
@Katherine_Scott talked with ZDNet about open source in IAC
IAC university teams use any software they want in the Dallara AV-21 racecars, open source or commercial without restriction. IAC started with 41 universities and 31 teams. Through a series of elimination challenges and team mergers, this was whittled down. All 9 IAC teams representing 21 universities that made it through to the final IAC race ran ROS 2 with Eclipse Cyclone DDS and Zenoh V2X (IAC how to here). Some used Autoware autonomous driving packages. Most teams ran Foxy but TUM made the move to Galactic Patch Release 1. TUM team leader Alexander Wischnewski said “Moving from ROS 2 Foxy to ROS 2 Galactic with default RMW gave us tremendous improvements.” Here is a deep dive into Galactic Patch Release 1 RMW performance.