Accelerating Collaboration and CV Deployment

Tampa (THEA) Connected Vehicle Pilots Works with the Infrastructure Integrator to improve Vehicle to Infrastructure (V2I) Connected Vehicle (CV) Applications

The Tampa Hillsborough Expressway Authority (THEA) Connected Vehicle (CV) Pilot project deployed 47 Roadside Units (RSUs) along THEA’s Reversible Express Lane (REL) and in Tampa’s Central Business District (CBD). At its peak, THEA deployed over 1,000 Onboard Units (OBUs) in personal vehicles, buses, and streetcars.

The infrastructure integrator was a THEA partner from the initial concept development phase and provided support to THEA during the development of the response to the USDOT’s Notice of Funding Opportunity (NOFO) for the CV Pilot project. RSUs were on their product roadmap prior to responding to the NOFO in support of THEA. The infrastructure integrator developed the THEA CV Pilot specific Vehicle to Infrastructure (V2I) applications as part of their product roadmap. The RSU hardware and wireless stack software baseline is developed and maintained by a global team for all international regions, while the V2I software applications, including THEA, New York City and Columbus are developed and maintained in Austin, Texas. The combination of having a local footprint and global development team allowed the infrastructure integrator to be very responsive to issues discovered during on-site testing. On several occasions, the global team, working in a different time zone, was able to resolve problems overnight and allow testing on-site to continue the very next morning, including an overnight turnaround of issues for the initial OmniAir RSU certification testing.

In addition to implementing basic V2I applications, the RSU was used to evaluate specialized applications. In these cases, the RSU operated correctly to identify unexpected application performance. The infrastructure integrator obtained the Multimodal Intelligent Traffic Signal Systems (MMITSS) software from the ITS CodeHub (formerly the OSADP) to integrate it to THEA’s CV Pilot deployment. As MMITSS was ported to run on their RSU, it became clear that the algorithm had been designed for CV penetration rates of 25% or more. After conferring with University of Arizona’s Dr. Larry Head, leader of the MMITSS project, the project team decided to add advance detection cameras to one corridor (Florida Ave). Such cameras were needed to collect sufficient vehicle data for the MMITSS queue length estimation algorithms to work properly. It later turned out that this estimation is incorrect due to algorithm assumptions. The infrastructure integrator worked with the Center for Urban Transportation Research (CUTR) to develop a queue estimation algorithm based on Basic Safety Messages (BSMs) locations and speeds. The team later learned that MMITSS does not feed this information to the Intelligent Traffic Signal System (I-SIG) to control signals and maximize traffic flows in real time. The infrastructure integrator is currently working with the City of Tampa to use video detection to supplement the low penetration rate of participant vehicles.

In another instance, the Transit Signal Priority (TSP) application within MMITSS was demonstrated to work correctly but did not provide a coordinated transit corridor. The infrastructure integrator worked with the City of Tampa to implement NTCIP 1211 standardized messages in the RSU to the signal controller to support the TSP messages between the RSU and transit vehicle. This NTCIP 1211 methodology supports a coordinated transit corridor using the standardized connected vehicle messages between the RSU and transit vehicle.

In addition, the infrastructure integrator implemented a version of the Pedestrian Mobility (PED-SIG) application. However, issues arose during early demonstrations that caused the application to be eliminated from the CV Pilot. This primarily had to do with the number of Wi-Fi networks at intersections as well as the inconsistent features of smartphones.

Finally, the infrastructure integrator developed a third V2I application, which was a version of the Pedestrian in a Signalized Crosswalk Warning (PED-X) application.  Due to smartphone GPS location inaccuracies, two LIDAR units were installed at the crosswalk. The smartphone location service could not reliably differentiate pedestrians walking on the sidewalk or in the street. To support the research goals for pedestrian safety, the RSU received pedestrian locations from the system integrated with the LIDAR. The RSU would then translate the vendor-specific protocol containing pedestrian locations to broadcast Pedestrian Safety Message (PSM)s over Dedicated Short-Range Communications (DSRC) for approaching CV-equipped vehicles to receive and process. After successful demonstration and several months of operation, one of the LIDAR units failed. When subsequent attempts to address the issue did not resolve the issue, a decision was made to replace the LIDAR units with thermal cameras. Though the RSUs were generally functioning correctly, one can conclude that the external devices needed to provide the data or interface to the RSU had challenges.

The most challenging applications to implement were Over-the-air (OTA) software updates as well as the data upload from the OBUs to the RSUs. There was no pre-existing application for either OBUs or RSUs. The team collaborated to create a unique set of applications that allowed data to be moved from the OBU to the RSU and vice versa in an efficient manner. However, because this was a never implemented as a CV application, there were many challenges in not only getting the application working reliably but getting correctly formatted data payloads. Some of the challenges included:

  • File size: while trying to minimize the file sizes, there were times when they could be large (e.g. over 100 megabytes).
  • Channel bandwidth: The bandwidth is limited and only three DSRC service channels were used (e.g. 176, 178, and 182).
  • Time vehicle in range: Depending on its location, a vehicle maybe in range for only a few seconds to connect and transfer data.

The team was able to overcome these challenges by developing a compression algorithm to compress the files to small payloads and dedicate RSUs to data and firmware transfer as well as tuning of the applications.
The following conclusions can be drawn from the THEA CV Pilot project team’s experiences:

  1. Adapting current ITS technology to create the necessary CV data needed by the applications was more challenging than was anticipated.
  2. Development of the OTA and data transfer application was technically challenging as it required large amounts of data to be transferred while vehicles were in motion traveling up to 65 miles per hour.  Because of the vehicle’s speed and small bandwidth available, it was difficult to successfully download data. 
  3. Critical applications required for system operation were underestimated or not anticipated at all, such as OTA and DataLogTransfer applications. However, these applications became the most essential to the operation of the pilot, while at the same time being the most complex to develop since they were not off-the-shelf.
  4. Specialized applications were improved through unanticipated solutions.
    1. Queue measurement using BSM speed and distance data instead of discharge
    2. LIDAR sensors for accurate pedestrian location instead of smart phone/Wi-Fi
    3. Video and thermal sensors to replace LIDAR for improved reliability and cost
    4. Video detection to supplement low BSM penetration for signal control
    5. Implementing NTCIP 1211 and SAE J2735 standards into multiple signal controller manufacturers to coordinated TSP corridors

November 2020