HMI warning strategy based on radar-video fusion perception for freeway merging areas: A risk field approach considering beyond-visual-range event.

Under beyond-visual-range conditions where the driver's sight is obstructed for the freeway merging area, the merging vehicle may seriously affect the mainline driver's safe passage. This study aims to evaluate the mitigation effect of radar-video fusion-based human-machine interface (rHMI) warnings on the interaction risk imposed on a mainline vehicle by a merging vehicle, addressing the risk warning under beyond-visual-range conditions. Three rHMI warning schemes were developed and tested through driving simulation. They are no radar-video fusion perception without rHMI warning (Baseline), partial coverage of radar-video fusion perception with one-level rHMI warning (Warning I), and full coverage of radar-video fusion perception with two-level rHMI warning (Warning II). The study constructed the kinetic energy field incorporating speed difference (KEF-SD). A convolution operation was used to extract the field strength's time-varying characteristics. The KEF-SD was compared with surrogate safety measures (SSMs) to demonstrate its performance. Results indicate: (1) rHMI warnings help mainline drivers to effectively reduce speed. The progressive intervention of two-level warning induces earlier speed adjustment; (2) The KEF-SD is reliable and valid. It can more accurately and finely explain the vehicle interaction process under beyond-visual-range conditions; (3) Compared to Baseline, Warning I and Warning II achieve maximum field strength reductions of 56.33 % and 89.38 %. Under the progressive intervention of two-level warning, the interaction risk mitigation is significant; (4) The provision and earlier supplementation of rHMI warnings stabilize field strength and reduce instantaneous risk mutations to zero. rHMI warnings may alleviate drivers' workload, and the two-level warning performs best; (5) It is recommended to prioritize full detection coverage of the merging area to maximize risk mitigation through staged rHMI warning for beyond-visual-range merging events. Results provide preliminary evidence for application of risk detection and warning in specific scenarios, enhancing safety in merging areas.