A distributed local Kalman consensus filter for traffic estimation

This work proposes a distributed local Kalman consensus filter (DLKCF) for large-scale multi-agent traffic density estimation. The switching mode model (SMM) is used to describe the traffic dynamics on a stretch of roadway, and the model dynamics are linear within each mode. The error dynamics of the proposed DLKCF is shown to be globally asymptotically stable (GAS) when all freeway sections switch between observable modes. For an unobservable section, we prove that the estimates given by the DLKCF are ultimately bounded. Numerical experiments are provided to show the asymptotic stability of the DLKCF for observable modes, and illustrate the effect of the DLKCF on promoting consensus among various local agents. Supplementary source code is available at https://github.com/yesun/DLKCFcdc2014.