Traffic Flow Estimation Using Cloud-Based Traffic Velocities for In-Vehicle Application

Increasing importance of automated driving and driver assistance systems are driving automotive industry in recent times. Research activities try to include information on the current traffic situation, for improved predictions or assistant functions, which emphasizes the need for more accurate traffic data. Cloud services already offer current traffic speed information, but it often is inaccurate or piecewise constant. Therefore traffic flow modeling presents a useful tool to reconstruct the traffic state using a distributed parameter traffic state estimation. Since typical traffic estimation algorithms are based on stationary detector data, this work proposes a new estimation setup, that incorporates a distributed measurement of less accurate, but cloud-based traffic velocities. It is shown that the designed Extended Kalman Filter can handle measurement disturbances in velocity and estimates the underlying traffic states, comprising velocity and density. The benefit of this approach is that only online live traffic data is used and thus the setup is made deployable for individual in-vehicle applications.