After initial testing and evaluation in Metro Atlanta, BlueMAC Bluetooth and ELSAG ALPR systems were deployed on a one-mile long, single-lane closure, freeway work zone on I-285 in September, October, and November 2012 and the iCone system was added to the evaluation in April 2013. Data collected from each of the three technologies was compared with manually collected travel time data. To evaluate the selected technologies, system-reported travel times were compared with travel time data obtained by manually matching vehicle license plates from overpass video cameras. While BlueMAC Bluetooth and iCone systems readily provide real-time travel time data via the service provider’s webpage, the ELSAG ALPR system stores license plate readings on the processing unit. The evaluation of the selected systems addresses the following two aspects: 1) accuracy of the reported travel time and 2) timeliness of data reporting.
Collect traffic and travel time data from all traffic lanes because travel times can vary substantially between lanes. Travel time differentials exist across freeway work zone travel lanes, which means placement of data collection technologies may bias collected speeds towards the specific lanes that are monitored or from which more data are collected due to the nature of the technology. The speed differential across lanes can introduce significant bias into monitored travel time results. It is important to set up equipment to capture representative traffic conditions when multiple lanes are involved and the speeds vary across lanes.
Specific examples of how this impacted the results include:
- ALPR travel times were longer because the system detected more vehicles in the slower (outside) lanes, which obstructed the view of vehicles in the inside lanes. The ALPR is biased toward capturing travel times of vehicles in the lane closest to the camera location. The iCone has a similar limitation because it has a low profile.
- Video data collected from overpasses were biased toward the faster moving lanes because large vehicles and trucks obstructed many license plates in the slower lanes.
- The iCone takes spot speed measurements and assumes the measurements are representative of conditions between those points, so long distances between devices could yield unreliable estimates.
- Slower-moving vehicles are more likely to be detected by Bluetooth as they are in the field of detection (300 ft radius from the sensor) for longer than fast moving vehicles. Though Bluetooth strategies a less impacted by placement because they are not visually gathering data.