1. Were any new/updated cyber-security mechanisms implemented to prevent or improve the security of the connected vehicles (CV)?
Please see the NYC CVPD documents at: https://www.its.dot.gov/pilots/events.htm, specifically the Security Management Operational Concept (SMOC).
2. Will the data collected from the Pilot be available for future studies and planning activities?
Pilot data can be downloaded from the USDOT’s ITS DataHub’s Connected Vehicle Pilot Sandbox: https://usdot-its-cvpilot-publicdata.s3.amazonaws.com/index.html.
3. Where can we obtain more technical details on the make/model of edge devices used, setup/configuration of edge devices, software on edge devices, etc.?
Please see the NYC CVPD documents at: https://www.its.dot.gov/pilots/events.htm, specifically the System Architecture Document (SAD) and System Design Document (SDD).
4. Did your analysis assess event data such as hard braking/acceleration and anti-lock brake systems (ABS)?
Hard braking/acceleration was only determined based on the trajectories and acceleration data within recorded BSMs around a CV application warning. The pedestrian in crosswalk warning (PEDINXWALK) application analysis assessed hard barking occurrences or events. The other CV application analyses did not classify hard braking events, although deceleration rates were part of assessing driver responses to CV app warning. Data regarding ABS systems or ABS activations were not available.
5. Are you supplying information to the lawyers for AASHTO/ITSA in support of their lawsuit against the FCC to get the NPRM in docket 19-138 overturned?
No, opinions on regulations are outside the scope of the project.
6. Regarding security, was the Security Credential Management System (SCMS) used to provide a security layer for Vehicle-to-Everything (V2X) communication?
Yes. Please see the NYC CVPD documents at: https://www.its.dot.gov/pilots/events.htm, specifically the Security Management Operational Concept (SMOC).
7. On Slide 17, what kind of location errors are these? Small errors or errors showing the vehicle at another street/location? Can you please identify the error type and its magnitude?
Location errors were identified in event logs where the location of the BSM triggered for the host vehicle was more than 30 meters from the centerline (by direction) of any public roadway while within city limits, or when the heading of the triggered BSM was not in agreement with any nearby roadway segment. Additional smaller and localized location drift or noise did remain in the obfuscated event records, but these smaller errors were not removed from consideration as location errors.
8. Did you capture front radar data (i.e., distance to lead vehicle)? Also, are some of the vehicles equipped with Adaptive Cruise Control (ACC)?
No to both questions.
9. Did the Apache Log4j hack impact any of your connected vehicle systems?
No.
10. Regarding Forward Collision Warning (FCW), how many warnings are being issued relative to the expected number of collisions, i.e., is the FCW threshold too low causing the system to generate false warnings?
The project did not evaluate the expected number of collisions. Driver response to the warnings was evaluated, not the efficacy of the applications.
11. Do drivers know that their vehicle is being monitored? How do you prevent that bias from the data?
Yes. The knowledge of monitoring bias may be present in all data and could not be prevented.
12. Can vehicles warn pedestrians crossing from the middle of the road instead of a designated crossing walkway?
The current thermal sensors are facing crosswalks. But yes, they can be configured to face a midblock, and it is possible to detect peds and warn the CV drivers.
13. Can you define what is considered a false alert?
A false alert designation was not fully adopted in the evaluation of the CV applications; determinations were only made if the driver responded with actions expected, given the CV warning application triggered. If referring to the Driver survey, the determination of a false alert was left to the judgement of the driver respondent.
14. How loud were the alerts?
The onboard units (OBUs) included both a speaker and microphone. The OBU had a Minimum Volume Config and Maximum Volume Config in percentages and not in Decibels (Example: 70% - Min, 90% - Max). The logic was that if the microphone picked up any background noise, it would play the warning at Maximum Volume, and if not, then it would play at Minimum Volume. Due to installation location of the microphones, there was variability in the field. Therefore, the default configuration was 70% volume.
15. Can you please comment on FCC prospects to favor 5G CV-X and kill DSRC going forward?
No, opinions on regulations are outside the scope of the project.
16. At what distance/time are alerts activated to avoid collision with bikes and pedestrians?
It varied with vehicle speed and deceleration characteristics.
17. On slide 42: How does the speed in work zones analysis take into consideration the vehicle types in the platoon -- say trucks/ tankers and other heavy vehicles?
The analysis focuses on the driver response to the alert and not surrounding traffic.
18. Are the thermal pedestrian detection videos archived and if so, can they be shared?
No to both questions.
19. For Red Light Violation Warning (RLVW), how many vehicles that reacted were moving in excess of the posted speed limit? Conversely, was the reaction similar to those vehicles that were at or under the posted speed limit?
The analysis focuses on the driver response to the alert and not surrounding traffic.
20. How did you measure the drivers' response/reaction to the warnings/alerts? As a change in the vehicle's state e.g., deacceleration? What about the drivers' perception and the assessment of the situation? Based on the driver's situational awareness, the driver may not react to the alert.
Please see the NYC CVPD documents at: https://www.its.dot.gov/pilots/events.htm, specifically the Performance Measurement and Evaluation Support Plan (PMESP) and the forthcoming System Performance/Evaluation Report.
21. What percentage of the pedestrian application was assessed during dusk and darkness?
For the field tests of the PED-SIG (PED) application for visually impaired pedestrians, approximately 80% of the tests were completed in daylight, with the remainder occurring in reduced light conditions (dusk to twilight conditions but nearby streetlights were on).
For the PEDINXWALK application, 90% of the events were recorded at times that would have been under dark sky conditions, but the nearby streetlights should have been on. The remainder would have occurred in twilight conditions.
22. Were false positive alert rates recorded for the applications?
This information is not directly recorded in the application or through any other in-vehicle recording of data, but during the data filtering and cleaning process. This involved removal of "invalid" alerts such as alerts with illogical event warning time scales. Driver responses were also evaluated to identify event conditions where drivers did not respond in a clear and logical way based on the CV application warning issued.
23. Can you please share the cost breakdown for this large-size deployment? What are the major items in terms of cost and their ratios?
Please refer to the below high-level cost table from the Comprehensive Deployment Plan:
24. Was the project able to achieve 1 meter accuracy with the location correction?
Tests showed that location accuracy was within the SAE J2945/1 accuracy requirements where additional location augmentation facilities were employed.
25. Is there any statistical data available on the participants in terms of their driving patterns, behavior, following rules, etc.?
No. Individual driver evaluations are outside the scope of the project and counter to the privacy needs of the project.
26. What is the density of Roadside Units (RSUs) in NYC?
On the North-South Avenues, RSUs are located at intersections on nominally 260 foot spacing.
27. Do you have a feel for what percentage of probable accidents were prevented by the deployment of your CV technology?
No.
28. A rental vehicle had each of the functions without a radio system. How will the data be used for insurance and road safety features?
Insurance programs are outside to scope of the NYC CV Pilot Deployment project evaluation.
Connected Vehicle Pilot Deployment and Program Evaluation
Phase 3 Operation and Maintenance (June 2019 - August 2022)
Phase 2 Design, Build, and Test (Sep 2016 - December 2020)
Phase 1 Concept Development (Sep 2015 - Sep 2016)
Preliminary Planning Phase (April 2014 - Sep 2015)
