The RAPSS sensing system consists of a stereo vision sensor, two 24GHz radar sensors and vehicle dynamics data. The task of the sensing system is to predict an imminent pedestrian impact and activate bumper and hood restraints before the impact. The design goal for bumper activation timing is 500ms before impact and for hood 200ms. Generally a vision system has good azimuth estimation and less good range estimation to targets. Radar sensors have good range estimation and less good azimuth estimation. A fusion of the vision and radar targets should provide good azimuth and range information. To be able to estimate an accurate activation timing, or Time To Impact (TTI), the range estimation needs to be accurate. Azimuth accuracy is needed to provide accurate impact prediction together with vehicle path prediction from vehicle dynamics. The RAPSS uses a fusion principle between vision and radar targets, i.e. for a fusion target to exist there needs to be both a vision detection and a radar detection. This implies that the fusion performance in terms of detection performance can not be better than detection performance from the vision system alone. This basic understanding and a need for cost reduction of the system led to the investigation of a vision system alone investigation where the radar sensors were removed.