NYC Connected Vehicle Pilot Found That 83 Percent of Participants Felt Safer When Using the Mobile Pedestrian Crossing App Compared to Not Using It.
Mobility and Safety Benefits and User Perceptions Were Evaluated for the New York City Connected Vehicle Pilot Deployment.
Made Public Date

New York City CV Pilot Deployment Results and Transition Plan

Summary Information

The New York City (NYC) Connected Vehicle (CV) pilot deployment was a large-scale CV technology deployment that primarily focused on safety applications that rely on vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and infrastructure-to-pedestrian (IVP) communications. The pilot's goals included improving safety by reducing vehicle and pedestrian crashes, injuries, and fatalities and improving mobility and reliability through crash prevention and lower crash severity. The project's performance results were shared in a public webinar after the completion of the project's operational phase.

The deployment, led by NYCDOT, included 470 roadside units (RSUs) and 3,000 City-fleet vehicles equipped with 6 V2V and 6 V2I applications. Additionally, two pedestrian applications were tested in the deployment. The V2I and V2V applications provided drivers with alerts so that the driver could take action to avoid a crash or reduce the severity of injuries or damage to vehicles and infrastructure.

One of the pedestrian applications, Mobile Pedestrian Signal System (PED-SIG), a custom smartphone application that provided the signalized intersection geometry conditions and the active pedestrian signals state, was field-tested with 24 participants with vision disabilities in November 2021. User perception and feedback on the PED-SIG were collected through pre-and post-experiment surveys.


The NYC CV pilot deployment collected 189,374 total events (defined as a warning, alert, or some other triggering parameter within the vehicle) from January 2021 to November 2021. All the event data was obfuscated, addressing privacy concerns. A control-treatment and before-after experiment design were used to evaluate the performance measures. Statistical analysis and surrogate safety measures (SSM, identifying specific surrogate indicators that are highly correlated to the risk of collisions) were used to assess the safety benefits of various applications through field-collected event data. Investigation of driver reaction to warnings, crash-based analysis, and SSM simulation analysis were also conducted. Driver perceptions and user experience with CV Apps were collected through three surveys: 1) Pre-deployment survey (last month of before period, 83 responses), 2) Early-deployment survey: (1-2 Months into after period, 19 responses), and 3) Late-deployment survey (4-5 months into after period, 161 responses).
For the PED-SIG app, 24 volunteer participants with vision disabilities were recruited via local and national organizations working with blind communities. Participants navigated four semi-protected signalized intersections using portable personal Pedestrian Information Devices (PIDs) accompanied by Institutional Review Board (IRB)-certified researchers. PED-SIG data were collected from operational data logs, field observation, and pre-and post-experiment surveys.


Safety Application Evaluation Results

  • Speed Compliance (SPDCOMP): Statistical analysis of the 40,635 SPDCOMP events showed that compared to silent warning scenarios, there were additional 47 events per 1,000 SPDCOMP events that drivers slowed to the speed limit when treatment was enabled. For driver responses, 0.148 m/s2 extra deceleration on average were observed after speed compliance warnings were issued.
  • Curve Speed Compliance (CSPDCOMP): Statistical analysis of the 27 CSPDCOMP events showed an 8.750 miles per hour (mph) reduction in vehicle speeds at curve entry, a 0.691 m/s2 reduction in lateral acceleration in the curve, and a 0.908 m/s2 decrease in deceleration difference.
  • Speed Compliance in Work Zone (SPDCOMPWZ): Statistical analysis of the 2,665 SPDCOMPWZ events revealed that there was an extra 0.427 m/s2 deceleration from the drivers on average after being issued the warnings. Moreover, 2.26 seconds reduction in time duration to slow down to speed limit were found.
  • V2V applications: Statistical analysis of the Intersection Movement Assist (IMA) application indicated a reduction of the conflict risk between host and remote vehicles after being given the IMA warning. Microsimulation results indicated a positive effect of Blind Spot Warning (BSW) and Lane Change Warning (LCW) applications in terms of reducing conflict risks.
  • Red Light Violation Warning (RLVW): Both statistical and simulation-based SSM analysis indicated a positive safety effect of the RLVW application. For example, drivers tended to decelerate approximately 0.137 m/s2 more after RLVWs were issued.
  • Pedestrian in Crosswalk (PEDINXWALK): The simulation results indicated a positive effect of PEDINXWALK application in terms of reducing conflict risks.

Driver Surveys

  • Thirty-eight (38) percent of drivers reported that the CV alerts helped them drive more safely.
  • The top three responses about concerns while using CV technology were distractions (reported by about 45 percent participants), false alerts or warnings (about 40 percent), and too many alerts or warnings (about 46 percent).
  • Thirty (30) percent of drivers reported that the CV alerts affected how they drive. Among these drivers, more than half (58 percent) reported the CV alerts affected how they drive positively or very positively.

PED-SIG Evaluation Results

  • Majority of the participants (83 percent) felt safer when using the mobile accessible pedestrian signal system (ped-sig) app compared to not using it.
  • All participants (100 percent) anticipated that pedestrians would benefit from the use of PED-SIG technologies.
  • Ninety-six (96) percent of the participants felt that they were given sufficient time to cross the intersection when using the PED-SIG application.
  • The main problems experienced during the field tests included inaccurate location information, slow responses, and inaccurate orientation
Results Type