An Improved Precrash Warning System with An Impact Energy-Based Estimator

Intersection collisions account for a significant portion of accidents and fatalities on roadways. We are developing a vehicle-to-vehicle cooperative collision warning system as part of our effort to enhance driver’s awareness of roadway situation and to reduce the accident rate. Central to the system is a threat assessment algorithm that is based on location and velocity information of the subject vehicle itself and those of its peers. Unfortunately, the location and velocity information from current positioning systems such as GPS is not very accurate. The uncertainties in the locations and velocity data leads to both missed warnings and false warnings. The paper first presents an analysis on how the warning criteria is affected by the uncertainties in the data. In the light of the analysis, we propose an impact-energy-based criterion to augment a previously reported collision warning logic, in order to reduce false alarms at small angle collision scenario.