A Method for Multi-Moving Target Detection and Tracking with 2D Laser Range Sensor
Masafumi Hashimoto, Yasuhisa Tanaka, Satoshi Ogata, Takeshi MURAYAMA, Fuminori OBA · TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series C · 2004
This paper presents a method for multi-moving target tracking with 2D laser range sensor (LRS) within indoor environment. Tracking candidates are extracted from LRS's output via heuristic rule and Interacting-multiple-model (IMM) estimator ; the heuristic rule extracts the forehand objects in the field of view of LRS while the IMM estimator does the curved line objects. Moving targets (i.e., moving curved-line objects) such as walking persons are found from the candidates, and they are tracked based on Kalman filter. An assignment algorithm in which a joint likelihood function is maximized provides the reliable data association related to multi-targets and multi-measurements in cluttered environments. A track masagement method is also presented to allow tracking multi-targets under their moving conditions such as temporal occlusion, crossing, merging and splitting conditions. An experimental result in which four walking persons were tracked in a corridor environment validates the proposed method.