Main Street Autonomy Enables MIT Driverless Autonomous Racing with Automatic Sensor Calibration Solution
PITTSBURGH, PA – September 24, 2024 – Main Street Autonomy, LLC (MSA), a robotics software and solution provider, announces a partnership with the MIT-PITT-RW autonomous racecar team, comprised of the Massachusetts Institute of Technology (MIT), the University of Pittsburgh (PITT), the Rochester Institute of Technology and the University of Waterloo (RW).
MIT-PITT-RW integrated the sensor calibration generated by MSA's Calibration Anywhere software into the team’s autonomous racecar (details online at https://driverless.mit.edu/). “The result is better than anything we’ve achieved ourselves,” said Andrew Saba, Engineer from Carnegie Mellon University Robotics Institute.
MSA's Calibration Anywhere software automates the cross-modal calibration process for perception sensors. Using only sensor data captured in any environment, the software generates 6DoF extrinsics relative to base_link for RGB and depth cameras, lidars, radars, IMUs, and GPS/GNSS antennas. It also calculates camera intrinsics and time offsets for all sensors.
The image above shows lidar-in-camera for several different lidars and cameras on the racecar. The image below is a lidar map of the Las Vegas Motor Speedway, built using motion-compensated lidar data, where motion compensation was performed using a fusion of visual odometry (from rolling shutter cameras) and other asynchronously-sampled sensor data.
“We are excited to work with the MITPITTRW team,” said William Sitch, Chief Business Officer at MSA. “Autonomous vehicles moving at such high speeds need absolutely perfect sensor calibration for seamless operation of the autonomy system. We have demonstrated our ability to deliver to this challenging requirement.”