An arterial speed estimation model fusing data from stationary and mobile sensors

This paper presents an arterial speed estimation model using data from two distinct sources: mobile probe vehicles and inductive loop detectors. The model consists of three modules: (1) the probe vehicle module which measures arterial speed using vehicles equipped with differential global positioning system receivers; (2) the loop detector module which estimates the link speed using loop detector data, incorporating traffic signal timing parameters; and (3) the data fusion module which uses a neural network to combine outputs from the above two modules to improve the speed estimation accuracy. The computational procedures of the three modules are presented. This paper presents a validation test of the model using a set of data generated from a calibrated simulation model. Our test results show that, the probe vehicle and loop detector modules are capable of making speed estimation with 2-RMSE of less than 3.20 km/h. Using a neural network to fuse the estimates from the two sources reduces the 2-RMSE to less than 1.32 km/h