Thorsten Sohnke

Reducing Uncertainties in Precrash-Sensing with Range Sensor Measurements

This paper focuses on the data processing development for Precrash- Sensing, when only the object range is provided. If the object is close to the sensor, perturbations in form of measurement uncertainty and other classes of noises will become critical due to the small amount of time available to make decisions. Nevertheless, these decisions are required to be accurate and highly reliable. The wide variety of object classes and possible situations in vehicle surroundings aggravate the effects of all these perturbations so that the signal processing development for Precrash becomes quite a challenge. Based on current techniques for object tracking and multilateration, crash-situations can be detected and crash-parameters such as velocity at the impact point can be computed. Experiments were performed with a vehicle equipped with a multiple-sensor system based on two different sensor technologies (radar and ultrasonic).