METHODOLOGY

The deployed vehicles experienced 107,609 alerts by V2I applications and 52,680 by V2V applications. Of all these events, 65,231 alerts were silent and 95,058 alerts were active. After data filtering and validation, the analysis broke down the events with valid alerts by treatment and control groups, before and after deployment periods, and silent and active alerts. The analysis identified and statistically described vehicle/driver response measures after alert onset for each application, including statistical tests to reveal any statistically significant differences in any of these measures between two experimental groups. Due to data limitation, the final safety assessment was achieved by comparing the response between all valid events with silent alerts and all valid events with active alerts, regardless of period (before or after) or vehicle group (treatment or control). The deployment was evaluated considering a “before” period from January 1 to May 19, 2021 and an “after” period from May 20 to December 31, 2021 using statistical comparisons between experimental groups.

FINDINGS

Based on the statistically significant difference in vehicle/driver response between events with silent and active alerts for each safety application, the key findings were summarized in Table 1.

Table 1. Statistical Results of Safety Impact Assessment of the NYC CVPD Applications.

The increase in speed limit compliance by SPDCOMP, reduction in red light violation rate by RLVW, reduction in near-crash rate in the LVD scenario by FCW, reduction in lane change rate by LCW, and reduction in unsafe lane change rate by LCW and BSW directly lead to potential safety benefits and the reduction in brake reaction time due to RLVW, FCW, EEBL, and IMA applications indirectly contributes to potential safety benefits from their deployment.