I-STREET Initiative – Evaluation of Intelligent School Zone Beacon and Vehicle-Cyclist Detection and Warning System
This study evaluated a smartphone-based application (app) with the capability to alert drivers when they exceed a given speed threshold in an active school zone. An additional component alerted drivers when they approached a cyclist and there was a collision risk. This vehicle-to-infrastructure (V2I) technology, tested in 2019 in Gainesville, FL, relies on intelligent school zone beacons that send signals to the servers connected to a smartphone-based app available to travelers for downloading. If the speed threshold that is set in the app is exceeded by a driver while within the school zone “geo fence”, an alert is triggered. A secondary service provided by this same app sends alerts about bicyclists (irrespective of whether they are in the school zone or not), which works provided that both the driver and the bicyclist have the app open.
- Modify school zone boundaries by direction and use high resolution GPS for higher lane-level precision to prevent false alerts. School zone boundaries are generally staggered in two directions (the point at which the driver sees end of school zone sign in one direction may not be the same place at which the start school zone sign in painted in the roadway in the opposite direction). The app evaluated in this study used the school zone marking on the roadway to define the geo fence. As a result, uneven sign placement sometimes led to false alerts.
- Synchronize data sources. The data recorded in this study was from various sources and the gaze data and GPS data were not synchronized. This precluded analyses such as how long it took drivers to slow down after they looked at school zone beacons or the cyclist.
- Keep the app upgraded to iron-out any functionality issues. While the overall performance of the app used in this study was generally robust, there were instances of the app being triggered even when the speeds did not exceed the specified threshold value (20 mph) and vice versa. This could be due to inaccuracies in GPS data, algorithms that trigger the alert, and/or latency issues, which should be kept in mind for app upgrades.
- Maintain a log of speeding alert trigger times. This would reduce manual data logging labor and also facilitate observation of changes in speeding behavior immediately after the alert is received.
- Be cautious about the suitability of attentional metrics to understand safe driving in school zones or safe driving around cyclists. Examining whether changes in gaze and saccades (fast eye movement from one fixation point to another) would translate into fewer crashes, less severe crashes, or fewer near-misses is an area identified for future work.