01001-机器人三维定位系统中关键技术的研究.pdfVIP

18 1 © Vol. 18 Suppl.1 2006 8 Journal of System Simulation Aug., 2006 , , , , , 116023 ICP ICP ICP TP391.9 A 1004-731X (2006) S1-0099-04 Investigation of Main Techniques in Robot Three-dimensional Localization GUO HeLI Han WANG Yu-xinJIA Qi LIU Tian-yang TANG Jun (Computer Department, Dalian University of Technology, Dalian 116023, China) Abstract A novel, laser-based approach for tracking the pose of a high-speed mobile robot was described. The approach has better performance in terms of computation time and accuracy, which is achieved by improvements of ICP algorithm. With filter, feature extraction and non-linear iterative techniques, the algorithm greatly decreases the running time while it remains a high accuracy. The algorithm also avoids the disadvantages of ICP and gets better accuracy and the possibility of converging to a local optimal value. Besides, the difficulty of gaining pitch information and the low accuracy of localization were solved by adding an obliquity sensor to the traditional robot localization system based on this novel scan matching algorithm. Key words three-dimensional localization; scan matching; feature extraction; ICP; linear filter 1 using Invariant Features