Primarily funded by the USDOT’s Smart City Challenge, the Smart Columbus Program is a collection of eight transportation, mobility and data studies aimed at improving access to jobs, enhancing tourism, stimulating the economy, connecting residents to safe and reliable transportation, and supporting efficient and sustainable movement of people and goods throughout Columbus. The Connected Vehicle Environment (CVE) study was one of the eight Smart Columbus studies with the goal to enhance safety and mobility for vehicle operators, both private and public, and to improve pedestrian safety in school zones. The CVE study deployed secure, high-speed wireless communication technology in 1,000 vehicles and at 85 signalized intersections along four district corridors to exchange critical situational data between the infrastructure and vehicles, and between individual vehicles. The collected data was then used to alert drivers of potential safety issues through a heads-up display via the following applications: Red Light Violation Warning, Reduced Speed School Zone Warning, Emergency Vehicle Preemption, Transit Signal Priority, Intersection Movement Assist, Lane Change Warning/Blind Spot Warning, Emergency Electronic Brake Light Warning, and Forward Collision Warning. The CVE launched in October 2020 and continued through May 2021.
The lessons learned from the deployment of the CVE are summarized based on stakeholder engagement with CVE, data collection, surveys of fleet drivers, and self-assessment based on performance metrics.
- Utilize a secure IPv6 connection to allow on-board unit (OBU) vendors to connect to their respective remote services. At the time of the study, no standards existed for performing firmware updates to OBUs, and thus each connected vehicle (CV) pilot site developed its own approach to performing the updates. By providing a secure IPv6 connection, any OBU vendor was able to perform firmware updates with a scalable and non-proprietary solution. Select anomalies in the firmware were addressed in this manner, such as reducing power draw in the non-operational state, and voltage levels for physical inputs/outputs.
- Perform regression testing of the full system any time that any component is changed. While unit testing and individual component testing is important, it did not relieve the need for a complete and thorough full system regression test any time that a system component was updated or modified to revalidate prior tests.
- Bin data before processing to reduce analysis run-time. The sheer volume of data collected from the CVE operational period resulted in lengthy query run-times. Binning the data beforehand can reduce the query run-time significantly. Enhancements to the operating system may also be required to support complex analysis for multi-step algorithms used to determine if in-vehicle warnings were indicated.
- Understand differences in vendor implementations of road-side units (RSU). Despite robust standards for RSUs, there were several subtle but important differences amongst vendor implementations, including differences in the deployment-critical Simple Network Management Protocol (SNMP) implementation, variance in the visibility of indicator bulbs on the exterior of the RSUs, and differences in how a Wave Service Advertisement (WSA) was implemented.
- Understand the time requirements for integration testing of the full set of CV applications. Despite being assessed as technology readiness level 6 or higher (early through advanced prototypes or full commercial availability), hundreds of hours of integration testing were required to fine-tune and validate the full set of CV applications. Primarily, inconsistent GPS accuracy due to local terrain and electromagnetic interference required repeated fine-tuning and testing of antennae to obtain consistent and satisfactory application performance.
- Utilize partnerships with private auto shops to perform sensor installations. Participants noted that they had a favorable opinion of Smart Columbus working with small-business auto shops to perform installations, indicating a willingness to travel up to 20 minutes to do so. Agencies should also provide a pick-up/drop-off service for participants to increase participation and willingness for installation.
- Continuously engage with stakeholders and maintain positive impressions through effort of working individuals. Continuous engagement with all stakeholders in the study helped to circumvent any disruptions to the schedule and plan, especially when contacts or team members working with the study changed. Maintaining a positive impression was important for this engagement, which was reflective of the specific individuals’ working on the study and their efforts.
- Run an integrated, multi-channel outreach campaign to recruit private drivers. A well-defined and robust private driver recruitment program was needed to achieve a sufficient level of private drivers for the study. This included a comprehensive email campaign to share information with participants, word-of-mouth tactics, and paid radio advertisements. Most private drivers that engaged with the study identified with messaging that targeted them as “early adopters”. Partnering with a trusted community organization helped allay concerns about privacy and security of information amongst potential drivers.