LADAR and Image Automatic Calibration for Improved Slam Performance

Lee, Cheng-Lung, Chaomin Luo, Mohan Krishnan, and Mark Paulik

Robots and autonomous vehicles are increasingly being considered for use in a wide variety of civilian and military applications such as fire fighting, enforcing perimeter security, bomb disposal, exploration, etc. Autonomous vehicles rely heavily on vision and Laser Detection and Ranging (LADAR) systems to provide mapping data for localization. The collected data are typically processed by a Simultaneous Localization and Mapping (SLAM) algorithm to iteratively build a map. In order to have reliable mapping data suitable for safe navigation, it is very important to register the camera image and LADAR data. Unfortunately, the fusion and calibration of the imaging data from these two systems often proves to be a computational bottleneck.

We have developed an efficient procedure to automatically register LADAR data with calibrated image data. The procedure includes: A) removal of tangential distortion from the camera image, B) using a calibration object to obtain a rotation matrix to register the camera image and LADAR data, and C) extracting image features for localization of objects of interest. The described algorithm provides a significant improvement in mapping accuracy and is expected to be used in the 2009 Intelligent Ground Vehicle Competition (IGVC).