Search

Deep reinforcement learning has found great successes in addressing many challenging control problems; however, real-world implementations are still scarce if not non-existent. This is primarily due to three main challenges pertaining to the implementation, the stability, the optimal settings, and a lack of knowledge on methods that can be applied to field settings. This research attempts to address these issues with an adaptive simulation-based control framework proposed specifically for the training and evaluation. The control framework has implemented simulation models to conduct an extensive sensitivity analysis on the effects of key design variables, including rewarding schemes, state spaces, and model training parameters. The feasibility of transfer learning as a training strategy is also studied on scenarios with different layouts and different driver behavior models. Complex scenarios are also evaluated and used as test cases, including multiphase ring-and-barrier control and multi-intersection control. The research has contributed a significant amount of evidence on several critical design and implementation-related questions such as input representation, data (technology) requirements, training methods, and model transferability.

This paper proposes a real-time winter road surface condition (RSC) monitoring solution that automatically generates descriptive RSC information in terms of snow and ice coverage by using images from fixed traffic and weather cameras. Several state-of-the-art pre-trained deep neural networks are customized and fine-tuned to address a specific domain, classifying the amount of snow coverage on a road surface. A thorough evaluation is conducted to identify and select the best model. This evaluation uses an extensive set of experiments to test the accuracy and generalization of each model and uses transfer-learning to fine-tune each of the pre-trained models on independent images from different traffic and weather cameras. The transferability of each model, relationship between model performance and data size, and the system settings of each model are then examined. Lastly, three online weight calibration methods are proposed to automatically update the model in new environments. The result shows that re-training the model using images from a mixed set of cameras has the most promising results.

This paper discusses a machine-learning traffic signal control method. A full-scale corridor is analyzed and the transferability of using a model pre-trained on a single intersection is examined. Two controller designs are explored, a simple two-phase design and a full ring-and-barrier style controller. The full ring-and-barrier controller adapts many of the key features present in traditional controllers, such as protected-permissive left turns, so that they can be used in the reinforcement learning (RL) paradigm. This study is the first to propose a method that uses deep reinforcement learning (DRL) to implement a full ring-and-barrier style controller. The study also examines the feasibility of using transfer learning to pre-train a model on a single intersection and then fine-tune it for application in a complete environment. Training is done on a simple four lane intersection and the pre-trained model is then transferred for fine-tuning to six controllers operating on a corridor modeled with field data obtained for University Avenue in Waterloo, Ontario, Canada. The performance of the fully trained model is then compared with the existing signal plans in relation to the average delay and average queue length. Application of the ring-and-barrier design to this corridor was found to reduce delays by at least 5% and average queue lengths at intersections by 27%.

Road safety performance function &#x0028 SPF &#x0029 analysis using data-driven and nonparametric methods, especially recent developed deep learning approaches, has gained increasing achievements. However, due to the learning mechanisms are hidden in a &#x201C black box &#x201D in deep learning, traffic features extraction and intelligent importance analysis are still unsolved and hard to generate. This paper focuses on this problem using a deciphered version of deep neural networks &#x0028 DNN &#x0029 , one of the most popular deep learning models. This approach builds on visualization, feature importance and sensitivity analysis, can evaluate the contributions of input variables on model &#x02BC s &#x201C black box &#x201D feature learning process and output decision. Firstly, a visual feature importance &#x0028 ViFI &#x0029 method that describes the importance of input features is proposed by adopting diagram and numerical-analysis. Secondly, by observing the change of weights using ViFI on unsupervised training and fine-tuning of DNN, the final contributions of input features are calculated according to importance equations for both steps that we proposed. Sequentially, a case study based on a road SPF analysis is demonstrated, using data collected from a major Canadian highway, Highway 401. The proposed method allows effective deciphering of the model &#x02BC s inner workings and allows the significant features to be identified and the bad features to be eliminated. Finally, the revised dataset is used in crash modeling and vehicle collision prediction, and the testing result verifies that the deciphered and revised model achieves state-of-the-art performance.

This research investigated the route choice behavior of cyclists in Toronto, Ontario, Canada, with data collected from a smartphone application deployed to many cyclists in the city. For the study, 4,556 cyclists registered and logged more than 30,000 commuting trips over 9 months. In addition to the time-stamped, second-by-second GPS readings on each trip, information on age, gender, and rider history was collected on a voluntary basis. Multinomial logit route choice models were estimated for the commuting cycling trips. The results revealed the critical importance of cycling facilities (e.g., bike lanes, cycling paths and trails) on cyclists’ route choice decisions, and provided valuable information for use in Toronto’s ongoing bicycle network planning.

Improving the quality of service of a rail transportation system, and enhancing the operation safety require a quantitative understanding of the dynamic and stochastic characteristics of its train operations, especially those caused by unexpected disruptions. In this paper, based on historical train operation records, the characteristics of the primary delays (PDs) occurred on the Wuhan-Guangzhou (WH-GZ) high-speed railway (HSR) are investigated, with a specific focus on the underlying behavioral and physical factors. Alternative distribution models, including Log- normal, Weibull, and Gamma distributions, are calibrated and subsequently tested using hold-out data, to investigate the temporal and spatial distributions of PDs. The Kolmogorov- Smirnov (K-S) test results show that all the candidate models can fit the PD distribution curves, however, the Log-normal distributional form outperforms the other models. Subsequently, the model validation, carried out on the test dataset and the entire data, supported by the results obtained from the K-S two-sample test, indicate that the Log-normal model could satisfy the requirements with sufficient accuracy. Keywords–High-speed railway; Primary delay; Spatial and temporal distribution; Kolmogorov-Smirnov test; Train operation records

Real-time winter road surface condition (RSC) monitoring is of critical importance for both winter road maintenance operators and the travelling public. Accurate and timely RSC information during snow events can help maintenance operators to deliver better maintenance services, such as plowing and salting, for reduced costs and salt usage and improved level of service. With this information, the traveling public can make more informed decisions on whether or not to travel, where to go, and which highways to drive on. In our previous effort we have shown the potential of applying a pre-trained convolutional neural network (CNN) for automatically detecting winter road surface conditions based on images from fixed traffic/weather cameras or in-vehicle devices. This paper focuses on comparing the performance of four most successful CNN models available from the leaders of this technology, namely, VGG16 (Oxford University), ResNet50 (Microsoft), Inception-V3 (Google) and Xception (Google), for solving the RSC classification problem. The models were first customized with additional fully-connected layers of neurons for learning the specific features of the RSC images. The extended models were then trained with a low learning rate for fine-tuning by using a small set of RSC images. The models were tested using a hold-out set of images from cameras installed at different locations, showing highly encouraging results.

This paper presents a study aimed at developing zonal-specific semivariograms for zones with different climates using regionalized random variables for a strategic road weather information ...

This paper presents the results of an extensive field study of the comparative performance of road salt on different pavement types for snow and ice control in transportation facilities. Approximately 400 tests were conducted in a real-world environment, covering three different pavement types and 27 snow events. The performance is compared on asphalt concrete (AC), portland cement concrete (PCC), and interlocked concrete (IC) pavements in terms of pavement clearing speed. The study suggests that on average, salt performs better on AC than PCC or ICC pavement, with the latter two having similar performance. The results were confirmed with a paired t-test analysis and then used to develop a performance model, the results of which were used to develop an adjustment factor for each of the different pavement types. The results from this research can be applied by pavement maintenance personnel to optimize salt usage and improve safety in transportation facilities.

This paper presents the results of an extensive field study of the comparative deicing performance of conventional dry road salt and prewetted road salt for the snow and ice control of transportation facilities. Approximately 500 tests in side-by-side sections were conducted in a real-world environment, covering a wide range of weather conditions in 31 snow events. This paper compares the snow melting performance of prewetted salt to that of regular road salt in terms of bare pavement regain time. The pair-wise study conducted confirmed that, despite the lower amount of sodium chloride (i.e., using approximately 20% less salt) in the prewetted salt, the two types of treatments performed similarly in terms of bare pavement regain time. A multiple linear regression analysis was performed to identify the factors that are associated with the relative performance of prewetted salt as compared to dry salt. The modeling results indicate that the performance advantage of the prewetting strategy decreases ...