This paper presents a novel strategy for the construction of dense three-dimensional environment models by combining images from a conventional camera and a range imager. Ro- bust data association is ?rst accomplished by exploiting the Scale Invariant Feature Transformation (SIFT) technique on the textured images. The two-dimensional feature locations are then identi?ed in the range images and the full 3D in- formation is in turn employed to calculate the relative regis- tration between consecutive camera poses. Finally, the infor- mation from the range images is combined with the newly ob- tained transformation to form a single model of the environ- ment. Validating results from a robot operating in a rescue scenario are presented. This is a typical scenario where the technique can be most useful as robot odometry, if present, is often highly unreliable and other means of locating the robot are necessary.
Main Research Area:
2006 IEE International Conference on Man-Machine Systems (ICoMMS 2006)