Laser Motion Detection and Hypothesis Tracking from a Mobile Platforms

Abstract A method,to detect and track moving objects with a laser range scanner from a mobile plat- form is presented. Detection and tracking is achieved by noticing differences in laser scan data over time, and maintaining hypotheses which reflect the chance of the changes being caused by moving objects. A general model of movement is employed to allow the tracking of multiple types of objects, from humans to other moving platforms, in one system. Pure odometry,data and scan matching is used to compensate for motion, and hypotheses are maintained using a Kalman filter to estimate the velocity of objects. The approach is em- ployed to track objects in a range of environ- ments, where cameras mounted on an active head are used to record the any movement around the platform.