RGB-D DE-based Scan Matching: Exploiting Colour Properties in Registration

Publication Type:
Journal Article
Citation:
Journal of Intelligent and Robotic Systems: Theory and Applications, 2015, 80 (1), pp. 71 - 85
Issue Date:
2015-10-07
Full metadata record
Files in This Item:
Filename Description Size
RGBD_DE_colour_ICP_online_published_paper_Dec14.pdfPublished Version1.43 MB
Unknown
© 2014, Springer Science+Business Media Dordrecht. Colour plays a fundamental role in the perception process of humans. In robotics, the exploitation of this type of information has become increasingly important in many different tasks. The development of new sensors has made it possible to obtain colour information together with depth information about the environment. We have recently developed a scan matching algorithm based on evolutionary concepts (Differential Evolution). The main objective of this work is to include colour properties in the registration process, studying how colour can be used to improve the scan matching process. In particular, we have designed a filter to extract the most significant points of a RGB-D scan based on the Delta E divergence between neighbours. In addition, colour properties have also been included in the fitness function of the scan matching method. Our approach has been tested in a real environment and the most significant conclusion is the improvement of the algorithm performance when measuring the valley of convergence.
Please use this identifier to cite or link to this item: