Seoungbum Kim

Measuring Freeway Traffic Conditions with Transit Vehicles

Many public transit agencies have equipped their fleet with automatic vehicle location (AVL) systems, which periodically provide the location of each vehicle in the fleet. Although the AVL is deployed for transit operations, the vehicles also provide valuable information about the traffic stream throughout the road network. This study developed a methodology to mine the transit AVL data to find all trips that use any portion of a prespecified freeway segment. These trips are then used to measure travel time and average speed over the freeway and thereby quantify conditions on the facility. The results are validated against concurrent loop detector data from a corridor. The greatest benefits, however, are in areas without fixed vehicle detection, so the methodology is also demonstrated on such a freeway corridor. The study corridors typically have fewer than 50 observations per day per kilometer per direction, so this paper includes a process for selecting those segments with at least one observation per hour. Even with this low density of observations, the data are aggregated to show clearly the recurring congestion patterns. Nonrecurring events are also evident, but they take longer to detect. With a higher frequency of observations (e.g., from other fleet AVL systems, cell phone tracking, or vehicle–infrastructure integration probe data), the methodology should also be effective for rapidly identifying nonrecurring congestion.

Evaluation of Axle-Based and Length-Based Vehicle Classification Stations

This study evaluated the performance of three permanent vehicle classification stations on freeways against concurrent video-based ground truth. All stations had dual loop detectors and a piezoelectric sensor in each lane, which together provided axle-based and length-based classification. Evaluation was done at individual, per vehicle resolution for each vehicle that passed during the study periods (more than 18,000 vehicles in uncongested conditions). Although the stations exhibited good performance overall (97% correct), the performance for trucks was poor; for example, only 60% of the single-unit trucks (SUTs) were correctly classified. All observed errors were diagnosed. Some errors could be fixed quickly, and others could not. Data from one site were used to revise the classifier to solve almost all fixable errors, and then performance at another location was tested. A chronic error found in the research was intrinsic to the vehicle fleet and may be impossible to correct with existing sensors: the shorter SUTs have a length range and axle spacing range that overlap those of passenger vehicles. Depending on calibration, SUTs may be counted as passenger vehicles or vice versa. Such errors should be expected at most classification stations. All subsequent uses of the classification data must accommodate this unavoidable blurring error. Because of the blurring, the axle classification station cannot be uncritically used to calibrate the boundary between passenger vehicles and SUTs for length-based classification stations, because unavoidable errors in axle-based classification would be amplified in the length-based classification scheme.

Multiple-Step Traffic Speed Forecasting Strategy for Winter Freeway Operations

Accurate and timely predictions of traffic conditions are required for congestion avoidance and route guidance in real-time freeway traffic operations. Special attention to winter operations is needed because prediction error could be amplified under severe weather conditions involving snow. This study employed a vehicle detection system to propose a speed prediction methodology that used the k–nearest neighbors algorithm. The speed prediction was further evaluated under different weather conditions with a road weather information system. Cross-comparisons of the mean absolute percentage error (MAPE) between three weather conditions (normal, light snow, and heavy snow) revealed that the MAPE tended to increase with increases in the forecasting time step (T) and snow intensity. The marginal MAPE over the time step was larger during heavy snow conditions than under normal and light snow conditions. These findings indicate that for winter freeway operations, the time step should be selected dynamically, depending on the weather conditions rather than with a static strategy for all conditions. To this end, this study proposes a framework to determine a dynamic forecasting T that is associated with weather conditions.

Assessing the performance of SpeedInfo radar traffic sensors

ABSTRACT Traffic speed is a crucial input for real-time traffic management applications. Operating agencies typically deploy their own sensors to collect the measurements, e.g., loop detectors. Recently, SpeedInfo emerged with a different paradigm for traffic speed collection: instead of selling hardware to operating agencies, at each link the company deploys its own Doppler radar in a self-contained wireless unit to measure traffic speeds and then sells the speed data. This study uses well-tuned loop detector-based speed measurements to evaluate 15 of the Doppler radar sensors over several months while the two traffic data collection systems were operating concurrently. The extended study period includes potentially challenging and transient conditions for the radar sensors: both recurrent (rush hour congestion and late night low flow) and nonrecurrent (incidents and precipitation). The analysis took a broad overview, comparing speed measurements from the radar sensors against the concurrent loop detector data and then explicitly looked for any anomalous pattern in the radar data such as latency and system outages. The work found the radar measurements are generally good, but also identified several points that should be considered before deployment, including latency, different biases in free flow and congestion, vulnerability to precipitation, and sensitivity to mounting angle.