Reliable, lane-level, absolute position determination for connected and automated vehicles (CAV’s) is near at hand due to advances in sensor and computing technology. These capabilities in conjunction with high-definition maps enable lane determination, per lane queue determination, and enhanced performance in applications. This project investigated, analyzed, and demonstrated these related technologies. Project contributions include: (1) Experimental analysis demonstrating that the USDOT Mapping tool achieves internal horizontal accuracy better than 0.2 meters (standard deviation); (2) Theoretical analysis of lane determination accuracy as a function of both distance from the lane centerline and positioning accuracy; (3) Experimental demonstration and analysis of lane determination along the Riverside Innovation Corridor showing that for a vehicle driven within 0.9 meters of the lane centerline, the correct lane is determined for over 90% of the samples; (4) Development of a VISSIM position error module to enable simulation analysis of lane determination and lane queue estimation as a function of positioning error; (5) Development of a lane-level intersection queue prediction algorithm; Simulation evaluation of lane determination accuracy which matched the theoretical analysis; and (6) Simulation evaluation of lane queue prediction accuracy as a function of both CAV penetration rate and positioning accuracy. Conclusions of the simulation analysis in item (6) are the following: First, when the penetration rate is fixed, higher queue length estimation error occurs as the position error increases. However, the disparity across different position error levels diminishes with the decrease of penetration rate. Second, as the penetration rate decreases, the queue length estimation error significantly increases under the same GNSS error level. The current methods that exist for queue length prediction only utilize vehicle position and a penetration rate estimate. These results motivate the need for new methods that more fully utilize the information available on CAVs (e.g., distance to vehicles in front, back, left, and right) to decrease the sensitivity to penetration rate.
