MSA- Visual SLAM in a multistory parking garage

This visualization follows the path of a robot building a map of the Bakery Square Parking Garage in Pittsburgh, PA. The simultaneous localization and mapping shown here use the output from four asynchronous cameras and an IMU mounted on the roof of a car. No lidar data or GPS/GNSS inputs are used.

The visualization demonstrates how MSA's Pose Engine generates 6DOF pose estimates (position + orientation) at high frequency with low compute latency. The online simultaneous localization and mapping (SLAM) algorithm is not expensive, consuming only one core of an Nvidia Jetson Xavier processor.

The green icons represent pose estimates, and each can be revisited with less than 1cm and 1mRadian accuracy after the robot has transited that portion of the map. With earth-centered earth-fixed (ECEF) coordinates, captured from an intermittent GPS receiver or known anchor points within the map, high ECEF accuracy is available as well.

Main Street Autonomy is a Pittsburgh-based software company providing calibration and localization/mapping software for on-road and off-road autonomy, both indoors and outdoors, on vehicles ranging from farm tractors, floor cleaning robots, tractor trailers, to drones. See our website for more information.

https://mainstreetautonomy.com/