When new technologies and methodologies are introduced, it is often hard for agencies to justify large-scale investments until there is confidence in the solution and a more complete understanding of the requirements of a proposed solution. Pilot deployments (typically a smaller and/or shorter-term deployment of a bigger solution) are great and allow an agency to confirm that a solution will provide the anticipated value to them, help them achieve their objectives, and help identify and allocate the appropriate resources to be successful.
While video-based traffic monitoring and proactive safety have been around for many years, in many aspects they are still relatively novel. Each of these two components has its own set of considerations when conducting a safety study. Luckily, once having experienced at least one deployment, these considerations will become very intuitive.
An example of where this was performed was in the PennDOT Smart Intersections Project. The very first site in this project was used to verify the process for site selection, data collection, video processing, and data analysis. The insights that were gained during the pilot ensured the remainder of the project was smooth sailing.
Site Information and Analysis
Washington Ave & Broad St, the site selected for the pilot deployment, is located in downtown Pennsylvania. It was selected due to its high volume (more than 1,000 pedestrians and dozens of cyclists use it daily). It is made up of two, two-way streets without any intersection skew, providing a representative mix of scenarios between motor vehicles and vulnerable road users.
Video data was collected and analyzed for the site over 5 consecutive days, yielding 120 hours of data. The results of the analysis indicated that around 30% of all close interactions were between vehicles and pedestrians. The movement that had the most interactions with the pedestrians was the southbound right-turning vehicle. Overall, among the network of intersections selected, this intersection had the highest pedestrian and cyclist conflicts, in terms of occurrences and rates. It was determined that this highly likely due to the high percentage of vulnerable road users.
Through this process, the team was able to adequately select the remaining sites to perform the analysis on. The following general insights can be gained from such a deployment that will ensure smooth sailing of any project, especially those that are of a large scale.
- Camera Setup Considerations: Obtaining the appropriate video footage for the location of intersect is the most important step in the process. Transoft Solutions always provides the client with a set of guidelines on the quality of cameras that are to be used, which is a one-and-done step; however, the actual setup can be more variable due to the physical characteristics of the site. The main factors that influence the setup are the number of cameras needed based on the size of the intersection and the physical infrastructure available to mount the camera. For example, in the case of the pilot site, all the poles available to mount the cameras on were too close to the intersection limiting the camera’s field of view, and not covering the entire intersection. This resulted in needing more than one camera at this location.
- Hours of Video Footage: One of the benefits of video-based proactive safety is that it requires a significantly shorter analysis period than traditional crash-based methods. For a standard study, Transoft recommends between 20 and 60, noncontinuous, hours of video footage, depending on the average volumes at the site. Additionally, in certain cases where the safety concern is more isolated, for example, if it is for a handful of movements rather than the whole intersection, additional hours of video may be beneficial. Regardless, the periods that the video footage is collected at can also be tailored to match the safety concern.
- Appropriate Site Selection: While this type of analysis can be used for any site and type of road segment, a deployment can be much easier and more efficient to perform at certain locations over others, especially if the budget is limited. This pertains to items such as road user volumes, physical infrastructure, and available knowledge on traffic/safety concerns at the location of interest, which affect the hours of video, camera setup, and analysis outputs, respectively. In the case of this pilot, the client obtained such a plethora of information that they decreased the pedestrian volume threshold to 100 pedestrians daily, significantly increasing the number of sites they can select from.
- Familiarity with Data: After just a single deployment, users of this system will have a much clearer and full-level understanding of the outputs they will be working with. This pertains to both the actual metrics themselves and how they will be displayed on the platform. Once that understanding is obtained, all future deployments can be tailored accordingly, and sites can be better selected based on the data that will be available. For example, after the pilot deployment was fully completed and the data analyzed, the client had decided that they would like to select a different desired design reaction time (used for conflict rates) than was initially used.
Upon the completion of the pilot deployment, all the insights were used to select the remainder of the 14 sites. The client took several months to ensure the selection of the appropriate site and the selection of the data collection time periods.
Lana Samara, B.Eng.
Senior Project Delivery Manager
Lana holds has a Bachelor’s degree in Civil Engineering from McGill University, specializing in Transportation Engineering. Lana works closely with public and private agencies around the world to define project scopes and to ensure results are delivered accurately and on time.
Lana has managed many deployments with road agencies, DOTs and consulting firms globally, including Boston, Seattle, Houston, Abu Dhabi, Ethiopia, Switzerland, The Hague, Bogota, and several more.