PAL Based Localization Using Pyramidal Lucas-Kanade Feature Tracker

The objective of the current research is to improve the localization accuracy of a mobile robot during its navigation in known environment. The robot uses GPS receiver to rough positioning. To determine more accurate location panoramic annual image of the omnidirectional imaging tool mounted on the CCExplorer robot is processed continuously. The applied software library algorithm [3] lets us compute optical flow based on the Lucas-Kanade feature tracker in real time. The optical flow computation is implemented in pyramidal fashion, from coarse to fine resolution. Consequently, the algorithm can handle large pixel flows, while keeping a relatively small window of integration, and therefore achieve high local accuracy (of the order of 0.1 pixel).
The paper summarizes the features of the PAL imaging tool, the applied computer vision method adapted to panoramic image and the results of the navigation experiment.