This paper presents a people tracker with a two-layered laser range sensor(LRS) and a fisheye camera set in an environment. The LRS detects the heads and knees of people; the camera takes color images of people detected by the LRS. From the position data of people taken by the LRS and the color histogram by the camera, heuristic-rule-based and global-nearest-neighbor based data association identify people. Their identified people are tracked via model-based tracker; the interacting-multi-model estimator and multi-model particle filter are applied to tracking people with sudden changes, such as walking, running, going/stopping suddenly, turning suddenly and jumping. Experimental result validates our people tracking method.