Robust Curb Detection from Multi-frame 2D Laser Range Finder Data  

作　　者：YUAN Xia SHI Peng-bo ZHAO Chun-xia 袁夏;史鹏波;赵春霞(School of Computer Science and Engineering,Nanjing University of Science and Technology)

机构地区：School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China

出　　处：《Journal of Donghua University(English Edition)》2016年第6期859-865,共7页东华大学学报（英文版）

基　　金：National Natural Science Foundation of China(No.61603184);the Key Laboratory of Intelligent Perception and Systems for High-Dimensional Information(Nanjing University of Science and Technology),Ministry of Education,China(No.30920140122006);the Fundamental Research Funds for the Central Universities,China(No.30915011321)

摘　　要：Curb detection is an important research topic in the unmanned ground robot. In this paper, a novel curb detection method using a 2D laser range fmder in an urban environment was presented. In the proposed method, firstly, 2D sequential laser range finder data were aligned using a Dead-Reckoning （DR） and GPS loeaHzation method based on extended Kalman filter. Secondly, curb points were extracted from multi.frame laser data based on spatial-temporal correlation analysis. Finally. the straight and curved curbs were detected by coordinate axis conversion least square linear fitting method and piecewise weighted least square fitting model respectively. The proposed method can detect both raised and sunken curbs robustly in campus environment. The proposed method has been verified in a real robot platform.Curb detection is an important research topic in the unmanned ground robot.In this paper,a novel curb detection method using a 2D laser range finder in an urban environment was presented.In the proposed method,firstly,2D sequential laser range finder data were aligned using a Dead-Reckoning(DR) and GPS localization method based on extended Kalman filter.Secondly,curb points were extracted from multi-frame laser data based on spatial-temporal correlation analysis.Finally,the straight and curved curbs were detected by coordinate axis conversion least square linear fitting method and piecewise weighted least square fitting model respectively.The proposed method can detect both raised and sunken curbs robustly in campus environment.The proposed method has been verified in a real robot platform.

关 键 词：curb dewction ROBOTICS range finder 

分 类 号：TP242.62[自动化与计算机技术—检测技术与自动化装置]