Xinkai Wu

Energy-Optimal Speed Control for Electric Vehicles on Signalized Arterials

Electrification of passenger vehicles has been viewed by many as a way to significantly reduce carbon emissions, operate vehicles more efficiently, and reduce oil dependence. Due to the potential benefits of electric vehicle (EV), many federal and local governments have allocated considerable funding and taken a number of legislative and regulatory steps to promote EV deployment and adoption. With this momentum, it is not difficult to see that in the near future, EVs could gain a significant market penetration, particularly in densely populated urban areas with systemic air quality problems. We will soon face one of the biggest challenges: how to improve the efficiency for the EV transportation system? This research aims to contribute to this field by proposing an analytical model that determines a time-dependent optimal velocity profile for an EV in order to minimize the electricity usage along a chosen route by systematically considering road characteristics and real-time traffic conditions. In particular, the proposed multistage optimal control model uniquely considers the impact of the presence of intersection queues in both temporal and spatial dimensions, which has been ignored in most traditional speed control models even for internal combustion engine vehicles. In addition, to facilitate the real-time operations, an approximation model, which simplifies the optimal speed profile, is further developed to increase the computation efficiency. The testing using the field data collected from a six-intersection signalized arterial corridor shows that the optimal velocity profile can significantly save energy for an EV, and the computational efficiency of the proposed approximation model is suitable for real-time applications.

Real-time estimation of arterial travel time under congested conditions

It is well-known that accurate estimation of arterial travel time on signalised arterials is not an easy task because of the periodic disruption on traffic flow by signal lights. It becomes even more difficult when the signal links are congested with long queues because under such situations the queue length cannot be estimated using the traditional cumulative input–output curves. In this article, we extend the virtual probe model previously proposed by the authors to estimate arterial travel time with congested links. Specifically, we introduce a new queue length estimation method that can handle long queues. The queue length defined in this article includes both the standing queue, i.e. the motionless stacked vehicles behind the stop line, and the moving queue, i.e. those vehicles joining the discharging traffic after the last vehicle in the standing queue starts to move. The moving queue concept is important for the virtual probe method because moving queue also influences the manoeuvre behaviour of a virtual probe. We show that, using the ‘event’ data (including both time-stamped signal phase changes and vehicle-detector actuations) collected from traffic signal systems, time-dependent queue length (including both standing queue and moving queue) can be derived by examining the changes in an advance detectors occupancy profile within a cycle. The effectiveness of the improved virtual probe model for estimating arterial travel time under congested conditions is demonstrated through a field study at an 11-intersection corridor along France Avenue in Minneapolis, MN.

Improving Queue Size Estimation for Minnesota’s Stratified Zone Metering Strategy

Stratified zone metering (SZM), a successor of zone metering, was recently implemented in the United States in the Twin Cities. SZM not only restricts metering rates subject to the freeway capacity constraints but also limits the on-ramp waiting time to a predetermined threshold. In SZM, accurate queue size estimation is crucial because inaccurate queue size can violate the maximum wait time constraint or reduce the service quality of the main-line system. Currently, a uniform and pre-calibrated regression equation is adopted for ramp queue size estimation for all ramps. It has been verified that such an estimation method may lead to outstanding estimation errors for ramp queues. In this paper, different ramp queue estimation algorithms are proposed for ramps with different categories, depending on the counting errors of queue and passage detectors. For ramps with minor counting error, queue size is estimated on the basis of flow conservation. For ramps with significant counting error on passage detectors but minor counting error on queue detectors, the flow conservation model can still be applied, but the traffic counts of passage detectors are replaced by the “green counts,” which are calculated by the ramp metering rate. For ramps with significant count errors from both queue and passage detectors, a Kalman filtering model is adopted. To verify the proposed methods, surveillance video data were used to compare actual and estimated queue sizes. Results indicate that the proposed methods greatly improve the accuracy of queue size estimation as compared with the original SZM queue size estimation.

Pedestrian Detection and Tracking from Low-Resolution Unmanned Aerial Vehicle Thermal Imagery.

Driven by the prominent thermal signature of humans and following the growing availability of unmanned aerial vehicles (UAVs), more and more research efforts have been focusing on the detection and tracking of pedestrians using thermal infrared images recorded from UAVs. However, pedestrian detection and tracking from the thermal images obtained from UAVs pose many challenges due to the low-resolution of imagery, platform motion, image instability and the relatively small size of the objects. This research tackles these challenges by proposing a pedestrian detection and tracking system. A two-stage blob-based approach is first developed for pedestrian detection. This approach first extracts pedestrian blobs using the regional gradient feature and geometric constraints filtering and then classifies the detected blobs by using a linear Support Vector Machine (SVM) with a hybrid descriptor, which sophisticatedly combines Histogram of Oriented Gradient (HOG) and Discrete Cosine Transform (DCT) features in order to achieve accurate detection. This research further proposes an approach for pedestrian tracking. This approach employs the feature tracker with the update of detected pedestrian location to track pedestrian objects from the registered videos and extracts the motion trajectory data. The proposed detection and tracking approaches have been evaluated by multiple different datasets, and the results illustrate the effectiveness of the proposed methods. This research is expected to significantly benefit many transportation applications, such as the multimodal traffic performance measure, pedestrian behavior study and pedestrian-vehicle crash analysis. Future work will focus on using fused thermal and visual images to further improve the detection efficiency and effectiveness.

A Hybrid Vehicle Detection Method Based on Viola-Jones and HOG + SVM from UAV Images

A new hybrid vehicle detection scheme which integrates the Viola-Jones (V-J) and linear SVM classifier with HOG feature (HOG + SVM) methods is proposed for vehicle detection from low-altitude unmanned aerial vehicle (UAV) images. As both V-J and HOG + SVM are sensitive to on-road vehicles’ in-plane rotation, the proposed scheme first adopts a roadway orientation adjustment method, which rotates each UAV image to align the roads with the horizontal direction so the original V-J or HOG + SVM method can be directly applied to achieve fast detection and high accuracy. To address the issue of descending detection speed for V-J and HOG + SVM, the proposed scheme further develops an adaptive switching strategy which sophistically integrates V-J and HOG + SVM methods based on their different descending trends of detection speed to improve detection efficiency. A comprehensive evaluation shows that the switching strategy, combined with the road orientation adjustment method, can significantly improve the efficiency and effectiveness of the vehicle detection from UAV images. The results also show that the proposed vehicle detection method is competitive compared with other existing vehicle detection methods. Furthermore, since the proposed vehicle detection method can be performed on videos captured from moving UAV platforms without the need of image registration or additional road database, it has great potentials of field applications. Future research will be focusing on expanding the current method for detecting other transportation modes such as buses, trucks, motors, bicycles, and pedestrians.

Perception of Waiting Time at Signalized Intersections

Perceived waiting time at signalized intersections differs from the actual waiting time and varies with signal design. The onerousness of delay depends on the conditions under which it is experienced. Using weighted travel time may contribute to optimal signal control if it can improve on the assumption that all time is weighted equally by users. This research explores the perception of waiting time at signalized intersections on the basis of the results of an online virtual experience stated preference survey. This survey directly collected the perceived waiting times and the user ratings of the signal designs of each intersection on an arterial that included three intersections. Statistically analyzing the survey data suggests that the perception of waiting time is a function of the actual time, and a quadratic model describes the relationship. The survey also indicates that there exists a trade-off between total waiting time and individual waiting time at each intersection. Drivers prefer to split the total waiting time across multiple intersections at the price of a longer total wait if the difference in the total waiting time of two signal designs is within 30 s. Survey data show that the perceived waiting time, instead of the actual waiting time, better explains how users rate the individual signal designs for intersections and arterials, including multiple intersections.

SMART-SIGNAL: Systematic Monitoring of Arterial Road Traffic Signals

Data collection and performance measurement for signalized arterial roads is an area of emerging focus in the United States. As indicated by the results of the 2005 Traffic Signal Operation Self-Assessment Survey, a majority of agencies involved in the operation and maintenance of traffic signal systems do not monitor or archive traffic system performance and thus have limited means to improve their operation. With the support from the Transportation Department of Hennepin County, Minnesota, we have successfully built a system for high resolution traffic signal data collection and arterial performance measurement. The system, named SMART-SIGNAL (Systematic Monitoring of Arterial Road Traffic Signals), is able to collect and archive event-based traffic signal data simultaneously at multiple intersections. Using the event-based traffic data, SMART-SIGNAL can generate time-dependent performance measures for both individual intersections and arterials including intersection queue length and arterial travel time. The SMART-SIGNAL system has been deployed at an 11-intersection corridor along France Avenue in Minneapolis, MN and the estimated performance measures for both intersection queue and arterial travel times are highly consistent with the observed data.

Estimation of red-light running frequency using high-resolution traffic and signal data

Red-light-running (RLR) emerges as a major cause that may lead to intersection-related crashes and endanger intersection safety. To reduce RLR violations, its critical to identify the influential factors associated with RLR and estimate RLR frequency. Without resorting to video camera recordings, this study investigates this important issue by utilizing high-resolution traffic and signal event data collected from loop detectors at five intersections on Trunk Highway 55, Minneapolis, MN. First, a simple method is proposed to identify RLR by fully utilizing the information obtained from stop bar detectors, downstream entrance detectors and advance detectors. Using 12 months of event data, a total of 6550 RLR cases were identified. According to a definition of RLR frequency as the conditional probability of RLR on a certain traffic or signal condition (veh/1000veh), the relationships between RLR frequency and some influential factors including arriving time at advance detector, approaching speed, headway, gap to the preceding vehicle on adjacent lane, cycle length, geometric characteristics and even snowing weather were empirically investigated. Statistical analysis shows good agreement with the traffic engineering practice, e.g., RLR is most likely to occur on weekdays during peak periods under large traffic demands and longer signal cycles, and a total of 95.24% RLR events occurred within the first 1.5s after the onset of red phase. The findings confirmed that vehicles tend to run the red light when they are close to intersection during phase transition, and the vehicles following the leading vehicle with short headways also likely run the red light. Last, a simplified nonlinear regression model is proposed to estimate RLR frequency based on the data from advance detector. The study is expected to helpbetter understand RLR occurrence and further contribute to the future improvement of intersection safety.

Car Detection from Low-Altitude UAV Imagery with the Faster R-CNN

UAV based traffic monitoring holds distinct advantages over traditional traffic sensors, such as loop detectors, as UAVs have higher mobility, wider field of view, and less impact on the observed traffic. For traffic monitoring from UAV images, the essential but challenging task is vehicle detection. This paper extends the framework of Faster R-CNN for car detection from low-altitude UAV imagery captured over signalized intersections. Experimental results show that Faster R-CNN can achieve promising car detection results compared with other methods. Our tests further demonstrate that Faster R-CNN is robust to illumination changes and cars’ in-plane rotation. Besides, the detection speed of Faster R-CNN is insensitive to the detection load, that is, the number of detected cars in a frame; therefore, the detection speed is almost constant for each frame. In addition, our tests show that Faster R-CNN holds great potential for parking lot car detection. This paper tries to guide the readers to choose the best vehicle detection framework according to their applications. Future research will be focusing on expanding the current framework to detect other transportation modes such as buses, trucks, motorcycles, and bicycles.

An Enhanced Viola-Jones Vehicle Detection Method From Unmanned Aerial Vehicles Imagery

This research develops an advanced vehicle detection method, which improves the original Viola-Jones (V-J) object detection scheme for better vehicle detections from low-altitude unmanned aerial vehicle (UAV) imagery. The original V-J method is sensitive to objects’ in-plane rotation, and therefore has difficulties in detecting vehicles with unknown orientations in UAV images. To address this issue, this research proposes a road orientation adjustment method, which rotates each UAV image once so that the roads and on-road vehicles on rotated images will be aligned with the horizontal direction and the V-J vehicle detector. Then, the original V-J can be directly applied to achieve better efficiency and accuracy. The enhanced V-J method is further applied for vehicle tracking. Testing results show that both vehicle detection and tracking methods are competitive compared with other existing methods. Future research will focus on expanding the current methods to detect other transport modes, such as buses, trucks, motorcycles, bicycles, and pedestrians.