Abstract
This paper describes a laser-based people tracker with two-layered laser range sensor. The upper-layered laser is scanned at waist height of people, and the lower-layered laser is done at knee height. A heuristic-rule based detection and the global-nearest-neighbor based data association find walking and running people from the laser images in crowded environments. Their people are tracked via the interacting-multiple-model estimator. Simulation and experimental results validate the effectiveness of our method.