As one type of object detection, small detection has been widely used in daily-life-related applications with many real-time requirements, such as autopilot and navigation. Although deep-learning-based methods have achieved great success recent years, they are not effective most them cannot achieve processing. Therefore, this paper proposes a single-stage network (SODNet) that integrates the specialized feature extraction information fusion techniques. An adaptively spatial parallel...

In recent years, online ride-hailing services have emerged as an important component of urban transportation system, which not only provide significant ease for residents' travel activities, but also shape new behavior and diversify mobility patterns. This study provides a thorough review machine-learning-based methodologies on-demand services. The importance in the spatio-temporal dynamics traffic is first highlighted, with macro-level research demonstrating its value guiding design,...

AI techniques, and data mining, with applications in complex real-world problems such as autonomous vehicles urban mobility.

Real-time traffic state (e.g., speed) prediction is an essential component for control and management in urban road network. How to build effective large-scale system a challenging but highly valuable problem. This study focuses on the construction of solution designed spatio-temporal data predict systems. In this study, we first summarize three challenges faced by prediction, i.e., scale, granularity, sparsity. Based domain knowledge engineering, propagation states along network...

In this study, we explore the problem of adaptive vehicle trajectory control for different risk levels. Firstly, introduce a sliding window-based car-following scenario extraction method, propose new alternative traffic conflict assessment metric, and build comprehensive library. Secondly, based on deep reinforcement learning (RL), design an algorithm, which is called <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Deep Adaptive Control</i> ,...

In current years, the improvement of deep learning has brought about tremendous changes: As a type unsupervised algorithm, generative adversarial networks (GANs) have been widely employed in various fields including transportation. This paper reviews development GANs and their applications transportation domain. Specifically, many adopted GAN variants for autonomous driving are classified demonstrated according to data generation, video trajectory prediction, security detection. To introduce...

Traffic flow prediction in spatio-temporal networks is a crucial aspect of Intelligent Transportation Systems (ITS). Existing traffic forecasting methods, particularly those utilizing graph neural networks, encounter limitations. When processing large-scale data, the depth these models can restrict their ability to effectively capture complex relationships and patterns. Additionally, methods often focus mainly on local neighborhood information, which limit capability recognize analyze global...

In intelligent transportation systems, one key challenge for managing ride-hailing services is the balancing of traffic supply and demand while meeting passenger needs within vehicle availability constraints. Accurate origin–destination (OD) predictions can empower platforms to execute timely reallocation cruising vehicles improve ride-sharing services. Nonetheless, complexity OD-based prediction arises from intricate spatiotemporal dependencies a higher need precision compared zone-based...

Recently, Heterogeneous Federated Learning (HtFL) has attracted attention due to its ability support heterogeneous models and data. To reduce the high communication cost of transmitting model parameters, a major challenge in HtFL, prototype-based HtFL methods are proposed solely share class representatives, a.k.a, prototypes, among clients while maintaining privacy clients’ models. However, these prototypes naively aggregated into global on server using weighted averaging, resulting...

The Autonomous Vehicle Storage and Retrieval System (AVS/RS) with tier-to-tier vehicles is modeled as a semi-open queueing network (SOQN). Different storage/retrieval requests in the AVS/RS are different classes of customers SOQN model. Analyzing multiple configurations an via computer simulation time-consuming. Therefore, this article uses some analytical methods to evaluate performances multi-class, multi-stage general service time interarrival distributions. Two synchronization policies...

Abstract How to effectively ensemble multiple models while leveraging the spatio‐temporal information is a challenging but practical problem. However, there no existing method explicitly designed for data. In this paper, fully convolutional model based on semantic segmentation technology proposed, termed as net. The proposed suitable grid‐based prediction in dense urban areas. Experiments demonstrate that through net, traffic state base can be combined improve accuracy.

Deep Neural Network (DNN) has been applied in a wide range of fields due to its exceptional predictive power. In this paper, we explore how use DNN solve the large-scale bus passenger flow prediction problem. Currently, most existing methods designed for problem are based on single view, which is insufficient capture dynamics fluctuation. Thus, analyze from scopes both macroscopic and microscopic levels, order take full advantage information variety views. To better understand role different...

How to effectively ensemble different base models is a challenging but extremely valuable task. This study focuses on the construction of an framework designed for spatio-temporal data predict large-scale online taxi-hailing demand, where attention-based deep net enhance prediction accuracy. We present three attention blocks model inter-channel relationship, inter-spatial relationship and position feature maps. Then, maps can be multiplied by input map adaptive refinement. The proposed...