On-road vehicle number (ORVN) is an important index used by Chinese traffic management departments to describe the total travel demand, evaluate the effectiveness of traffic signal optimization, and guide the design of vehicle restriction policies. One common approach to measure ORVN is to count the total amount of unique vehicles captured by roadside cameras during a certain amount of time. However, our study shows that to maintain this approach’s accuracy requires dynamic adjustment of observation period length, which makes this approach lack of feasibility in real world and its results lack of robustness. This paper proposes a new approach to estimate ORVN based on vehicle trajectory observations according to the fundamental diagram of traffic flow, as well as a corresponding strategy to deploy roadside cameras in road network for better estimation accuracy. Simulation results show our method achieved significantly improvement in ORVN estimation accuracy and reliability.

暂无