Dynamic Risk Detection and Early Warning Strategy for Vehicle Collisions at Signal-free Intersections*

This paper presents a dynamic risk detection and early warning strategy for vehicle collisions at signal-free intersections. Firstly, a joint collision risk detection method is proposed based on time to collision (TTC) and time exposure TTC (TET). The trajectory model and circular-bicircular model are innovatively used to detect and judge the collision of turning vehicles while redundantly processing uncertain factors. Secondly, based on the results of conflict detection, a two-level early warning mechanism based on TTC and TET for signal-free intersections is proposed. In the first-level early warning, a cooperative game model and a genetic algorithm determine and trigger the alerted vehicle. In the second-level early warning, the driver’s aggressiveness can be judged by identifying the driver’s speed change after the first-level early warning and different intensity early warnings can be triggered. Finally, based on the SUMO simulation platform, the test scenario of a signal-free intersection is constructed to test and analyze the effectiveness and applicability of our early warning strategy. The test results show that the proposed strategy can accurately identify all collision events and trigger early warning with a 100% success rate. The false alarm rate of early warning is 6.115%, and the average closest distance of false alarm cases is 4.209 m. Although there is no collision, the danger level at the intersection is quite high, so the proposed early warning strategy is very credible. Under different proportions of aggressive drivers, the proposed warning strategy can significantly reduce the collision rate and average collision kinetic energy.