Implementation of lane-by-lane detection at actuated controlled intersections

Previous work with lane-by-lane detection has shown efficiency gains during periods of moderate to low volume, as well as a decrease in overall intersection delay when lane-by-lane detection is compared with traditional movement-based detection. This research extends that work with the implementation of lane-by-lane detection at the Indiana Department of Transportation's test bed intersection in Noblesville, Indiana. Over a 3-week period with traditional movement-based detection, green interval lengths, volume-to-capacity (v/c) ratios, and cycle lengths were computed for all eight phases. Detection and controller settings were changed to implement lane-by-lane detection on the multilane movements at the test bed, and the same measures of effectiveness were calculated for 3 weeks of lane-by-lane detection. Dornig free operation with low to moderate volume, statistically significant decreases in green durations and cycle times, as well as statistically significant increases in v/c ratios, were observed on movements with lane-by-lane detection. These implementation results corroborate previous simulation results.