Measures of Effectiveness and Validation Guidance for Adaptive Signal Control Technologies
Chapter 6. Future Research and Development
This project has resulted in an open-source, web-based system for ingesting data from four sources of signal performance data and computing performance metrics and comparing performance for different conditions in order to validate that ASCT systems meet their operational objectives. The system produces comparative results, but does not generate conclusions about validity. Human analysts are still a necessary component of the process in many ways to process the data, interpret the findings, and generate reports. At minimum, future research and development is needed to identify higher-level quantitative metrics that map directly to certain operational objectives for direct validation, similar to the way HCM methods classify system performance on an A-F scale.
For example, the access equity objective generally requires that the green time allocated for each phase is appropriate for the demand—that the movement is provided enough green to process all vehicles waiting during the red phase, and not too much more. The red time for each phase is not unduly long, and particularly during red, the waiting vehicles do not perceive that other green time is being wasted when no vehicles are being serviced through the intersection. Using the GOR measure, for example, this qualitative goal might manifest itself as a metric to provide GOR values for all phases at the intersection as close to 80% as possible. Other target percentages might be selected. Other researchers (Zheng, et al., 2012) have suggested that perhaps an appropriate target for access equity is Webster splits.
For smooth flow operation, the objective is principally to drive the number of stops on the arterial route as close to zero as possible. Objective targets for the number of stops per mile and the average speed for various facility types need to be established similar to the way the HCM methods establish arterial performance from estimates of the various parameters. Research and development is needed to determine if the real-time, calculated components for platoon ratio and the other inputs to the HCM delay equations can be used for this purpose. In addition, metrics might be established from percent arrivals on green such as the following: “if the percent arrivals on green on an arterial route are greater than 75%, 80% of the time, then a smooth flow objective is validated”. Some areas of the country denote this kind of quantitative objective in a qualitative way as “maximizing the greens to reds ratio”, which is equivalent to driving the number of stops on the route as low as possible.
For queue management, the missing piece is obviously the measurement of queue lengths. Techniques are available in the literature (Liu, et al., 2009; 2011) to leverage advance detection zones for this. It is recommended that this method be incorporated or other similar methods be developed into the MOE calculation engines. Once these calculations are available, higher level quantitative metrics will be required that identify a traffic control system’s ability to meet this objective.
As an open-source project, we hope that researchers will extend the system in various ways noted above, and in ways we have not envisioned, and pass these improvements back to the open-source project for the betterment of all. The project is posted on the SourceForge web site.