Where am I? Place recognition for robots.
Our upcoming article in Robotics and Autonomous Systems describes a method by which a mobile robot can determine, based on laser range data, if it has visited a location before; and - if so - can orient itself within the existing map. This is the robot's equivalent of you being left blindfolded in an unknown location within a city and looking around to determine where you are so that you can find your way home. The technology can also be used to efficiently determine if a loop has been closed while incrementally building a map, or to merge maps generated at different times or by multiple vehicles.
Image by Mike Bosse
This work details the algorithms that have enabled us to build accurate maps using data collected from trajectories over hundreds of kilometers, and to localize vehicles in an industrial environment over thousands of hours of operation.
Michael Bosse and Robert Zlot. "Keypoint design and evaluation for place recognition in 2D lidar maps." Robotics and Autonomous Systems. Volume 57, Issue 12, 31 December 2009, Pages 1211-1224.
The paper is available here
Read more about our research in mapping and localisation.
- A map constructed from 2D laser data collected from 165km of driving on city streets in Brisbane, Australia overlaid on a Google Maps image. Two SICK LMS291 lasers are mounted horizontally on the roof of the vehicle and provide all of the data required to generate the map (i.e. no GPS, odometry, or inertial sensors are required).
Contact:
