An object-oriented data fusion system for mobile vehicle

This paper presents a multi-sensor fusion module embedded in a mobile vehicle. The problem considered here is the dynamic reconstruction of the vehicle's environment based on measurements of a set of sensors. The vehicle's environment is composed of moving and/or stationary obstacles. A typical problem of such a task is that a measurement of an obstacle may be inaccurate, incomplete or corrupted by noise. Another problem is the environment's dynamic nature, obstacles may enter or exit the visual field of the sensors, they may be temporarily occulted by other objects and false alarms may occur during a few time steps. To solve these problems, the system must be able to integrate information from successive measurements and evaluate temporal and spatial correspondence across multiple time steps or images. Dynamic environment modeling of a mobile vehicle can be viewed as a target tracking problem if one considers a moving observer (the mobile vehicle) and moving or stationary targets.