Probability Model of Time to Collision Considering Sensor Accuracy

The intelligent vehicle anti-collision system can effectively slow down or avoid the occurrence of traffic accidents and improve the driving safety of the vehicle. The time to collision (TTC) algorithm is not sufficient in anti-collision algorithm when the accuracy characteristics of different sensors are ignored. Therefore, this paper proposes an improved TTC threshold calculation method considering the influence of sensor detection error. In view of the influence of sensor detection error, a TTC probability model is constructed. In this model, the host vehicle (HV) speed, the target vehicle (TV) speed and the relative distance between them are regarded as a random process, and their distribution functions are established. Subsequently, these parameter distribution functions are used to construct the TTC probability model that approximates the normal distribution by Geary-Hinkley transformation. This model uses the collision probability index to calculate more reasonable TTC threshold. Experimental results were produced and comparative analyses were conducted for different cases. The obtained results confirm the effectiveness of the proposed TTC probability model.