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. Key words: deep reinforcement learning, traffic signal control, optimal control, 1. Introduction In recent years, breakthrough developments in the area of deep learning have transformed the artificial intelligence (AI) community. Today, computers are able to exceed the performance of humans [...]

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. Key words: road surface condition, realtime recognition, deep learning, convolutional neural network. Dans ce document, on propose une solution de surveillance en temps reel de l'etat de la surface de roulement (ESR) en hiver qui genere automatiquement des informations descriptives de l'ESR en matiere de couverture de neige et glace en utilisant des images de cameras fixes de circulation et meteo. Plusieurs reseaux de neurones profonds preentraines de pointe sont adaptes et peaufines pour traiter un domaine specifique, soit celui de classifier la quantite de couverture de neige sur une surface de roulement. Une evaluation approfondie est effectuee afin de determiner et de choisir le meilleur modele. Cette evaluation utilise un vaste ensemble d'experiences pour verifier la precision et la generalisation de chaque modele et utilise l'apprentissage par transfert pour peaufiner chacun des modeles pre-entraines sur des images independantes provenant de differentes cameras de circulation et meteo. La transferabilite de chaque modele, la relation entre la performance du modele et la taille des donnees ainsi que les parametres du systeme de chaque modele sont ensuite examines. Enfin, trois methodes de calibration de poids en ligne sont proposees pour mettre a jour automatiquement le modele dans de nouveaux environnements. Le resultat montre que le reentrainement du modele a l'aide d'images provenant d'un ensemble de cameras fournit les resultats les plus prometteurs. [Traduit par la Redaction] Mots-cles: etat de la surface de roulement, reconnaissance en temps reel, apprentissage profond, reseau neuronal convolutif., 1. Introduction In countries with severe winter seasons such as Canada, road surface conditions (RSC) on highways during snow events could vary remarkably from location to location and over time. [...]

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. Key words: winter maintenance, salting, pavement types, asphalt, interlocking concrete, sidewalks, environment. Dans le cadre de cet article, on presente les resultats d'une vaste etude sur le terrain au sujet du rendement comparatif du sel de voirie employe sur differents types de revetement pour le controle de la glace et de la neige sur les reseaux de transports. Environ 400 essais ont ete effectues dans un environnement concret, comprenant trois differents types de revetement et 27 evenements de chute de neige. On compare le rendement des revetements en beton asphaltique (BA), en beton portland (BP), et autobloquant en beton (AB), et ce, en fonction de la vitesse du deneigement et deglacage du revetement. L'etude suggere qu'en moyenne, le sel donne un meilleur rendement sur le revetement en BA que sur ceux en BP ou AB, et le rendement de ces deux derniers se rapproche. On a confirme les resultats fi l'aide de l'analyse du test t d'echantillons apparies qu'on a ensuite utilisee pour developper un modele de rendement, dont les resultats ont servi a determiner un facteur d'ajustement pour chacun des differents types de revetement. Les resultats de cette recherche peuvent servir au personnel de l'entretien des revetements en vue d'optimiser l'utilisation du sel et d'ameliorer la securite sur les reseaux de transport. [Traduit par la Redaction] Mots-cles : entretien hivernal, salage, types de revetement, asphalte, autobloquant en beton, trottoirs, environnemental., Introduction Safe and efficient pavement surfaces are essential for surface transportation system mobility and efficiency. To keep the surface transportation system safe and efficient in winter, large amounts of snow [...]