Forwards: A Map-Free Intersection Collision-Warning System for All Road Patterns

Collision warning is one of the most important functions of a vehicle safety system. The emergence and expansion of the applications of positioning techniques and dedicated short-range communication (DSRC) have promoted the collision warning system evolution from a simple ranging-sensor-based system to a cooperative system. Differing from prior work that relied heavily on the e-Map, high-accuracy differential Global Positioning System (DGPS), or advanced car features like the controller area network (CAN) bus, this paper proposes Forwards, i.e., a map-free intersection collision-warning system for all road patterns with lower requirement and lower cost accessories. Forwards employs a triple Kalman filter (tri-KF)-based estimator that integrates GPS and external inertial sensor measurement to provide calibrated motion state information (MSI) such as position, velocity, and acceleration of the vehicle. Each vehicle then adaptively broadcasts its own MSI via the DSRC-based protocol. Using the steady-state maneuvering model, short-term trajectories of local and neighboring vehicles are further predicted, based on their current MSI. Collision-detection algorithms are then designed based on the model of finding the minimum distance of vehicles' future trajectories, and hierarchical warnings are given upon different criteria. Simulation results show that our approach outperforms the referenced approach in successful warning ratio and requires far fewer accessories and external conditions than the other referenced approaches.