Predicting vehicle travel times by modeling heterogeneous influences between arterial roads

A system for predicting travel time of a vehicle on one or more routes within a network of the arterial roads, the system comprising:

• a memory with a set of instructions;
• a processor communicatively coupled with the memory;
• a plurality of probe vehicle data collection module configured to collect information of trajectories of one or more probe vehicles in a periodic fashion using GPS technology over a period of time, wherein the one or more probe vehicles are plying around the network of the arterial roads;
• a neighborhood structure data collection module configured to collect information of neighborhood structure for each link of one or more links of the network of the arterial roads, wherein the information of neighborhood structure for the link includes downstream links and upstream links, wherein the information of neighborhood structure includes a traversal time of the one or more probe vehicles in each link of one or more links of the network of the arterial roads, and wherein the each link of the network is assumed to be in any one of a congested state or a uncongested state;
• a spatio-temporal capturing module configured to capture spatio-temporal dependencies between the one or more links of the network of the arterial roads based on the neighborhood structure using a NoisyOR conditional probability distribution, wherein the spatio-temporal dependencies includes a congestion correlation based on the collected information of neighborhood structure for each link of the network of the arterial roads;
• a learning module configured to learn one or more parameters of a data driven probabilistic model from the collected information of the trajectories of the one or more probe vehicles traversed within the network of the arterial roads and captured spatio-temporal dependencies between each link of the network of the arterial roads using expectation maximization method;
• and
• a prediction module configured to predict travel time of the vehicle on the one or more routes within the network of the arterial roads using a real time information of trajectories of the one or more probe vehicles and the one or more learnt parameters of the data driven probabilistic model.