Travel time prediction is a hot research topic in ITS area, many researches focus at prediction of travel time on either freeway or arterial road network. The methodologies of these researches are highly dependent on the type of traffic data collected. In this Chapter, we depict the categories of traffic data collecting tools and then discuss the literature of travel time prediction.
2.1. Traffic Probing Tools
The probing tools can be used for measuring traffic data in two ways [10]: (1) logging the passage of vehicles from selected points along a road section or route that we regarded as site-based, or (2) using moving observation platforms traveling in the traffic stream itself and recording information about their progress, which we classify to vehicle-based. Concerning the site-based method, which includes registration plate matching, remote or indirect tracking, and input output methods and so on. The stationary observer techniques that include loop detectors, transponders, radio beacons, video surveillance, etc. In the past, many ITS studies and transportation agencies use the traffic data from dual-loop detectors which are readily available in many locales of freeways and urban roadways [10]. Dual-loop detector systems are capable of archiving with traffic count (the number of vehicles that pass over the detector in that period of time), velocity, and occupancy (the fraction of time that vehicles are detected). These records can be used for further traffic statistic research. On the other hand, the development and application of Radio Frequency Identification (RFID) might be extended to the real-time goods tracking in freight transport and the TTP issue in the
near future. In another way, the advanced registration plate matching techniques consist of collecting vehicles license plate and arrival times at various checkpoints, matching the license plates between consecutive checkpoints, and computing travel times from the difference between arrival times. Such as Automatic Vehicle Identification (AVI) method can recognize the license plate by video and transform it into digital data for later research. In addition, the cellular telephone systems are one of the potential techniques to provide travel time.
In group (2), the moving observer methods (vehicle-based) include the floating car, volunteer driver and probe vehicle methods are developed incrementally by collecting traffic dataset in recently years. The micro computer instrumentations (such as OBU) are designed and installed on vehicles to record vehicle speed, travel times, directions or distance it passed. Additionally, mobile data such as GPS is useful, and the GPS-GIS combination can contribute the efficiency in both data collection and results analysis [23], especially for volunteer driver and fleets of probe vehicles.
However, there exists no traffic information collection methodology can solve the above problems. For example, site-based TTP methods have the spatial coverage problem because the sensors or AVI devices are fixed and limited to obtain the real-time traffic data, and vehicle-based TTP methods have the cost and temporal coverage problems because the cost of probing vehicles is very high if a dedicated fleet of probing vehicle is maintained. In this thesis, we propose an LBS-based method which is vehicle-based. And the commercial fleets we used in experiment are taxi fleets equipped with LBS to record the real-time traffic data, and we regarded them as our probing tools.
2.2. Travel Time Prediction
There are numerous methodologies of TTP had been proposed in previous works, which can be categorized as follows [10]: regression methods (mathematics model), time series estimation methods, hybrid of data fusion or combinative models [21] and artificial intelligence method like neural network [14] Most of past studies estimate the travel time based on historical traffic data. In [16], Auto Regression (AR) model and state space model for time series modeling were used to predict travel time. The Kalman Filtering provides an efficient computational (recursive) in many TTP researches [2, 11, 23], because this filter is very powerful in several aspects: it supports estimations of past, present, and even future states, and it can do so even when the precise nature of the modeled system is unknown. In [22], the Support Vector Regression (SVR) model was used to predict travel time for highway users. [1] presented the pattern matching technique in TTP. For example, the traffic patterns similar to the current traffic are searched among the historical patterns, and the closest matched patterns are used to extrapolate the present traffic condition. [3] developed an OD estimation method to make more accurate estimation of traffic flow and traffic volume in congestion traffic status. Moreover, the data fusion models of TTP integrated grey theory [19] and neural network-based. [23] developed some hybrid models toward data treatment and data fusion for traffic detector data on freeway.
Besides, some artificial intelligence methods were applied to solve TTP issue.
PARAMICS of a real-world freeway section model [14] was proposed to develop an artificial neural network (ANN). [18] studied genetic algorithm to optimize performance of TTP. However, most of existing researches predicted travel time based on historical traffic data analysis and lost the precision because of disturbance of real-time events, such
as accidents, construction, signal break, and traffic block. In other words, most studies have been shown that prediction accuracy was often compromised by the underlying mechanism of prediction methods more than other influencing factors [4]. To extend the application of travel time information in open environments, such as arterial roads, the overcome of current difficulty is necessary [7]. For instance, signals and intersections are the main factors to influence prediction accuracy in arterial road sections of urban network. In this thesis, we propose a knowledge-based method with data mining technique to discover the spatiotemporal traffic rules and patterns from LBS-based applications, and also consider the intersection delay, links traffic conditions, weather, traffic events, and road geometry (attributes/interferences of TTP) to construct knowledge classes for solving the travel time issue.