INTERNATIONAL JOURNAL OF PRECISION ENGINEERING AND MANUFACTURING Vol. 14, No. 9, pp. 1-8 SEPTEMBER 2013 / 1 © KSPE and Springer 2013 Scan Matching Online Cell Decomposition for Coverage Path Planning in an Unknown Environment Batsaikhan Dugarjav 1 , Soon-Geul Lee 1,# , Donghan Kim 2 , Jong Hyeong Kim 3 and Nak Young Chong 1,4 1 School of Mechanical Engineering, Kyung Hee University, Seocheon-dong, Giheung-gu, Yongin, 446-701, South Korea 2 School of Electronic Engineering, Kyung Hee University, Seocheon-dong, Giheung-gu, Yongin, 446-701, South Korea 3 Dept. of Mechanical System Design Eng., Seoul National University of Science & Technology, Gongneung-dong, Nowon-ku, Seoul, 139-743, South Korea 4 School of Information Science, Japan Advanced Institute of Science and Technology, Asahidai, Nomi, Ishikawa 923-1292, Japan # Corresponding Author / E-mail: sglee@khu.ac.kr, TEL: +82-31-201-2506, FAX: +82-31-201-2506 KEYWORDS: Scan matching, Sensor-based online incremental cell decomposition, Oriented rectilinear decomposition complete coverage, Path planning This paper presents a novel sensor-based online coverage path-planning algorithm that guarantees the complete coverage of an unknown rectilinear workspace for the task of a mobile robot. The proposed algorithm divides the workspace of the robot into cells at each scan sample. This division can be classified as an exact cell decomposition method, which incrementally constructs cell decomposition while the robot covers an unknown workspace. To guarantee complete coverage, a closed map representation based on a feature extraction that consists of a set of line segments called critical edges is proposed. In this algorithm, cell boundaries are formed by extended critical edges, which are the sensed partial contours of walls and objects in the workspace. The robot uses a laser scanner to sense the critical edges. Sensor measurement is sampled twice in each cell. Scan matching is performed to merge map information between the reference scan and the current scan. At each scan sample, a two-direction oriented rectilinear decomposition is achieved in the workspace and presented by a closed map representation. The construction order of the cells is very important in this incremental cell decomposition algorithm. To choose the next target cell from candidate cells, the robot checks for redundancy in the planned path and for possible positions of the ending points of the current cell. The key point of the algorithm is memorizing the covered space to define the next target cell from possible cells. The path generation within the defined cell is determined to minimize the number of turns, which is the main factor in saving time during the coverage. Therefore, the cell’s long boundary should be chosen as the main path of the robot. This algorithm is verified by an experiment under the LABVIEW environment. Manuscript received: August 31, 2012 / Accepted: April 18, 2013 1. Introduction The task of covering a bound region of space is common to numerous applications. For example, cleaning robots are designed to automatically clean indoor workspaces. Cell decomposition is often employed in solving coverage problems. In this method, a target position must be reached without colliding with obstacles while the robot is moving. The main idea of cell decomposition is to decompose a given bound workspace into a set of non-overlapping regions. Each region is termed cell. The combination of these regions covers or approximates a subset of interest from the workspace, namely, the regions not occupied by obstacles. In the early research of cellular decomposition, a popular technique that yields a complete coverage path solution is the trapezoidal decomposition. 1 In this technique, the robot’s free space is decomposed into trapezoidal cells. The coverage for each cell can be easily achieved with simple back-and-forth motions because each cell is a trapezoid. Unfortunately, the trapezoidal approach has too much redundancy to guarantee complete coverage. To overcome this problem, there was introduced the boustrophedon decomposition, which was an enhancement of the trapezoidal decomposition. 2 This method was designed to minimize the number of excess lengthwise motions by merging narrow cells into one cell. Another development of the cell decomposition technique is the Morse decomposition algorithm presented by Acar, 3 where the location of cell boundaries is indicated by using Morse functions. Instead of analyzing the vertices of the given workspace, Acar’s algorithm looks for connectivity changes of the slice in the free space to locate the cell boundaries. These decomposition methods are formed by sweeping a line over the known workspace. A new cell boundary is created whenever the sweep line encounters a vertex. The performance analysis of the coverage path planning based on cell decomposition is investigated, 4 where the algorithm converts DOI: 10.1007/s12541-XXX-XXXX-X