An Adaptive Control Strategy with Parameter Estimation on Traffic Flow Model

X.-H. Yu (USA)


Parameter estimation, Markov decision control, traffic signal control, maximum likelihood estimation, adaptive control.


The main goal in traffic management is to set signals at traffic intersections to minimize the queue length and/or vehicle delay time. It has been shown that a traffic signal control problem can be formulated as a decision making problem for a stochastic dynamical system. To apply the Markov decision control theory to find the optimal signal settings at intersections, vehicle arrival rates (i.e., number of vehicles arrived at the intersection for each movement per hour) should be known in advance. When this information is not available (either unknown or it may fluctuate around the nominal value), an on-line parameter identification algorithm must be developed. In this paper, the maximum likelihood algorithm is employed for arrival rate estimation in a Markovian decision control process for traffic signals.

Important Links:

Go Back