<rss version="2.0" xmlns:atom="https://www.w3.org/2005/Atom">
  <channel>
    <title>Transport Research International Documentation (TRID)</title>
    <link>https://trid.trb.org/</link>
    <atom:link href="https://trid.trb.org/Record/RSS?s=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" rel="self" type="application/rss+xml" />
    <description></description>
    <language>en-us</language>
    <copyright>Copyright © 2026. National Academy of Sciences. All rights reserved.</copyright>
    <docs>http://blogs.law.harvard.edu/tech/rss</docs>
    <managingEditor>tris-trb@nas.edu (Bill McLeod)</managingEditor>
    <webMaster>tris-trb@nas.edu (Bill McLeod)</webMaster>
    <image>
      <title>Transport Research International Documentation (TRID)</title>
      <url>https://trid.trb.org/Images/PageHeader-wTitle.jpg</url>
      <link>https://trid.trb.org/</link>
    </image>
    <item>
      <title>A full-sample longitudinal trajectory reconstruction framework for signalized intersection approach lanes in a vehicle-infrastructure cooperation environment</title>
      <link>https://trid.trb.org/View/2762379</link>
      <description><![CDATA[Vehicle trajectories contain abundant spatial–temporal information that can be utilized for traffic state estimation and parameter optimization. In the vehicle-infrastructure cooperation environment, connected automated vehicles (CAVs) can provide their trajectory data and perceive the trajectory data of nearby vehicles within a specific range. Moreover, these trajectory data can be shared and cross-validated with fixed detectors, offering a novel solution for reconstructing full-sample vehicle trajectories at the signalized intersection. Based on the car-following (CF) model, this paper proposes a trajectory reconstruction framework that integrates CAVs sensing data and camera detection data. This framework can solve the trajectory reconstruction problem for regular vehicles (RVs) between any two consecutive CAVs on the lane. It comprises four key modules: the trajectory reconstruction model (TRM) based on the trailing CAV (TTRM), the TRM based on the leading CAV (LTRM), the Modified LTRM, and the fusion algorithm (FA). Furthermore, NGSIM and simulation data are utilized to validate and analyze the performance of the proposed framework under various CAV penetration rates (PRs) and traffic demands. Results indicate that: (1) compared to the state-of-the-art CF-based algorithms, the proposed framework significantly reduces in terms of position error, queuing position error, moving position error, time error, speed error, and acceleration error with average RMSE reductions of 83.45%, 20.97%, 90.87%, 85.54%, 30.39%, and 28.64% in NGSIM data, respectively, and even more substantial decrease in simulation data. (2) The framework maintains acceptable error rates across different PRs of CAVs and varying traffic volumes. (3) The framework is robust for different CF model parameters in simulation data. The findings can support signalized intersection management and control in the vehicle-infrastructure cooperation environment.]]></description>
      <pubDate>Thu, 10 Sep 2026 16:58:35 GMT</pubDate>
      <guid>https://trid.trb.org/View/2762379</guid>
    </item>
    <item>
      <title>Multi-Agent Deep Reinforcement Learning for Safe Autonomous Driving With RICS-Assisted MEC</title>
      <link>https://trid.trb.org/View/2672832</link>
      <description><![CDATA[Environment sensing and fusion via onboard sensors are envisioned to be widely applied in future autonomous driving networks. This paper considers a vehicular system with multiple self-driving vehicles that is assisted by multi-access edge computing (MEC), where image data collected by the sensors is offloaded from cellular vehicles to the MEC server using vehicle-to-infrastructure (V2I) links. Sensory data can also be shared among surrounding vehicles via vehicle-to-vehicle (V2V) communication links. To improve spectrum utilization, the V2V links may reuse the same frequency spectrum as the V2I links, which may cause severe interference. To tackle this issue, we leverage reconfigurable intelligent computational surfaces (RICSs) to jointly enable V2I reflective links and mitigate interference appearing at the V2V links. Considering the limitations of traditional algorithms in addressing this problem, such as the assumption of quasi-static channel state information, which restricts their ability to adapt to dynamic environmental changes and leads to poor performance under frequently varying channel conditions, in this paper, we formulate the problem at hand as a Markov game. Our novel formulation is applied to time-varying channels subject to multi-user interference and introduces a collaborative learning mechanism among users. The considered optimization problem is solved via a driving safety-enabled multi-agent deep reinforcement learning (DS-MADRL) approach that capitalizes on the RICS presence. Our extensive numerical investigations showcase that the proposed reinforcement learning approach achieves faster convergence and significant enhancements in both data rate and driving safety, as compared to various state-of-the-art benchmarks.]]></description>
      <pubDate>Tue, 01 Sep 2026 14:02:31 GMT</pubDate>
      <guid>https://trid.trb.org/View/2672832</guid>
    </item>
    <item>
      <title>Design and Optimization of a Hybrid VLC/THz Infrastructure-to-Vehicle Communication System for Intelligent Transportation</title>
      <link>https://trid.trb.org/View/2672830</link>
      <description><![CDATA[This paper proposes a hybrid infrastructure-to-vehicle (I2V) communication framework to support future 6G-enabled intelligent transportation systems (ITS) in smart cities. Leveraging existing LED streetlighting infrastructure, the system simultaneously delivers energy-efficient illumination and high-speed wireless connectivity. The proposed scheme integrates visible light communication (VLC) with a complementary terahertz (THz) antenna array to overcome VLC limitations under high ambient light and adverse weather conditions. Key contributions include the design of a VLC/THz access network, seamless integration with lighting infrastructure, a proposed switching-combination (PSC) mechanism, and a physical layout optimization strategy. Using a grid search method, thousands of configurations were evaluated to maximize lighting coverage, received power, signal-to-noise ratio (SNR), signal-to-interference-and-noise ratio (SINR), and minimize outage probability. Results show that optimized lighting coverage improves from 35% to 97%, while hybrid communication coverage increases from 49% to 99.9% at the same power level. Under extreme environmental conditions, the hybrid system maintains up to 99% coverage, compared to 69% with VLC alone. These results demonstrate the scalability, cost-efficiency, and practicality of the proposed system for next-generation ITS deployment.]]></description>
      <pubDate>Tue, 01 Sep 2026 14:02:31 GMT</pubDate>
      <guid>https://trid.trb.org/View/2672830</guid>
    </item>
    <item>
      <title>Optimization for Dynamic Multi-RIS-assisted SWIPT-Enabled V2I Networks: A Deep Learning Approach</title>
      <link>https://trid.trb.org/View/2622052</link>
      <description><![CDATA[Reconfigurable intelligent surfaces (RISs) have emerged as a highly promising technology in sixth-generation (6G) vehicular systems, offering the ability to dynamically control the wireless propagation environment. In this paper, we examine simultaneous wireless information and power transfer (SWIPT) by employing multiple RISs within a vehicle-to-infrastructure (V2I) communication system. The wireless environment exhibits high complexity due to fading and shadowing effects. To model this accurately, we adopt the double generalized Gamma (dGG) distribution. This comprehensive modeling approach enables a more realistic and insightful performance evaluation of RIS-assisted SWIPT systems under practical mobility and fading conditions. To reflect real-world vehicular dynamics, we incorporate a statistical Random Waypoint (RWP) mobility model, while also accounting for imperfections in channel state information (CSI) that arise due to high mobility and channel estimation errors. The study also integrates a non-linear energy harvesting (NL-EH) scheme to enhance performance via the power-splitting (PS) protocol. A unified objective function is proposed to jointly optimize transmit power and PS factors, aiming to maximize both the harvested energy and information rate. To address the non-convex nature of the problem, an iterative algorithm is utilized, supported by closed-form solutions derived from the Karush-Kuhn-Tucker (KKT) conditions and joint optimization (JO) method. Monte-Carlo simulations are conducted to verify the accuracy of the analytical results. Additionally, a deep neural network (DNN) framework is introduced for optimized value prediction, demonstrating superior SWIPT performance compared to single RIS configurations, with reduced complexity and faster execution.]]></description>
      <pubDate>Tue, 01 Sep 2026 14:02:30 GMT</pubDate>
      <guid>https://trid.trb.org/View/2622052</guid>
    </item>
    <item>
      <title>Edge-Assisted Ml-Aided Uncertainty-Aware Vehicle Collision Avoidance at Urban Intersections</title>
      <link>https://trid.trb.org/View/2709444</link>
      <description><![CDATA[Intersection crossing represents one of the most dangerous sections of the road infrastructure and Connected Vehicles (CVs) can serve as a revolutionary solution to the problem. In this work, we present a novel framework that detects preemptively collisions at urban crossroads, exploiting the Multi-access Edge Computing (MEC) platform of 5G networks. At the MEC, an Intersection Manager (IM) collects information from both vehicles and the road infrastructure to create a holistic view of the area of interest. Based on the historical data collected, the IM leverages the capabilities of an encoder-decoder recurrent neural network to predict, with high accuracy, the future vehicles' trajectories. As, however, accuracy is not a sufficient measure of how much we can trust a model, trajectory predictions are additionally associated with a measure of uncertainty towards confident collision forecasting and avoidance. Hence, contrary to any other approach in the state of the art, an uncertainty-aware collision prediction framework is developed that is shown to detect well in advance (and with high reliability) if two vehicles are on a collision course. Subsequently, collision detection triggers a number of alarms that signal the colliding vehicles to brake. Under real-world settings, thanks to the preemptive capabilities of the proposed approach, all the simulated imminent dangers are averted.]]></description>
      <pubDate>Mon, 31 Aug 2026 10:31:30 GMT</pubDate>
      <guid>https://trid.trb.org/View/2709444</guid>
    </item>
    <item>
      <title>Multi-Modal Sensing and Fusion in mmwave Beamforming for Connected Vehicles: A Transformer Based Framework</title>
      <link>https://trid.trb.org/View/2761415</link>
      <description><![CDATA[Millimeter wave (mmWave) communication, utilizing beamforming techniques to address the inherent path loss limitation, is considered as one of the key technologies to support ever increasing high throughput and low latency demands of connected vehicles. However, adopting standard defined beamforming approach in highly dynamic vehicular environments often incurs high beam training overheads and reduction in the available airtime for communications, which is mainly due to exchanging pilot signals and exhaustive beam measurements. To this end, we present a multi-modal sensing and fusion learning framework as a potential alternative solution to reduce such overheads. In this framework, we first extract the representative features from the sensing modalities by modality specific encoders, then, utilize multi-head cross-modal attention to learn dependencies and correlations between different modalities, and subsequently fuse the multimodal features to obtain predicted top-k beams so that the best line-of-sight links can be proactively established. To show the generalizability of the proposed framework, we perform a comprehensive experiment in four different vehicle-to-infrastructure (V2I) and vehicle-to-vehicle (V2V) scenarios from real world multimodal and 60 GHz mmWave wireless sensing data. The experiment reveals that the proposed framework (i) achieves up to 96.72% accuracy on predicting top-15 beams correctly, (ii) incurs roughly 0.77 dB average power loss, and (iii) improves the overall latency and beam searching space overheads by 86.81% and 76.56% respectively for top-15 beams compared to standard defined approach.]]></description>
      <pubDate>Fri, 28 Aug 2026 13:33:42 GMT</pubDate>
      <guid>https://trid.trb.org/View/2761415</guid>
    </item>
    <item>
      <title>Random Access for Semantic Transmission under Finite Buffer and Retransmission in Vehicular Networks</title>
      <link>https://trid.trb.org/View/2761473</link>
      <description><![CDATA[The rapid advancement of autonomous driving brings large amount of data to transmit in vehicular networks, which increases the burden of channels. While semantic communication (SemCom) can alleviate bandwidth pressure by transmitting semantics rather than raw data, random access (RA) for SemCom may still suffer from collisions and delays due to resource competition. This indicates that, in current vehicular networks, a unified modeling and analysis framework integrating SemCom and access schemes is supposed to be designed and explored. In this correspondence, we propose a RA scheme named SVRA under the buffer and retransmission constraints, where the state characterizations are determined using Markov chain. Furthermore, we introduce the age of incorrect information (AoII) by considering the structure similarity index measure (SSIM). We compare SVRA with existing schemes, the results of which validate that SVRA has a better performance.]]></description>
      <pubDate>Thu, 27 Aug 2026 10:04:05 GMT</pubDate>
      <guid>https://trid.trb.org/View/2761473</guid>
    </item>
    <item>
      <title>An Efficient, Identifiable and Abortable Multi-Party Signature Scheme for VANETs</title>
      <link>https://trid.trb.org/View/2761457</link>
      <description><![CDATA[Vehicular ad-hoc networks (VANETs) are networks based on short-range wireless communication technology, mainly used for direct communication between vehicles and interaction with roadside infrastructures. The emergence of VANETs has improved the efficiency and safety of vehicle travel. However, malicious vehicles may intentionally send incorrect messages to mislead other vehicles for personal gain, and this behavior cannot be identified yet. To address this issue, we propose an efficient, identifiable, and abortable multi-party signature scheme for VANETs. Specifically, we utilize the property of zero-knowledge proofs to design a method that can efficiently identify malicious vehicles during the signature process, and define a strategy to quickly locate false proofs. Additionally, the message signing key in the proposed scheme is jointly generated by multiple entities, which avoids the security issues of key escrow and single point failure. Through rigorous security analysis, it is demonstrated that the proposed scheme satisfies essential security requirements for VANETs, including message unforgeability and authentication, vehicle anonymity, malicious traceability, collusion resistance, replay attack resistance, identifiable abort, and forward security of signing private key. Performance analysis shows that our scheme reduces the total aggregation verification time by up to 80% compared to existing schemes, and maintains a lower packet loss rate (< 0.6%) at higher vehicle densities. Therefore, our scheme is suitable for large-scale VANETs deployments.]]></description>
      <pubDate>Thu, 27 Aug 2026 10:04:05 GMT</pubDate>
      <guid>https://trid.trb.org/View/2761457</guid>
    </item>
    <item>
      <title>Temporal and Feature Alignment in V2I Fusion for 3D Object Detection and Tracking at Intersections</title>
      <link>https://trid.trb.org/View/2767157</link>
      <description><![CDATA[Vehicle-to-infrastructure (V2I) cooperative perception is important for safer intersection operation because roadside sensors can reveal road users that are occluded, distant, or poorly observable from onboard cameras alone. Dependable V2I fusion is hindered by two practical issues. First, sensing, processing, and wireless transmission delays introduce temporal misalignment between infrastructure and vehicle streams. Second, large viewpoint differences create feature-level domain discrepancy that degrades cross-view matching and fusion. This paper proposes a unified alignment framework with a feature alignment module (FAM) and a temporal alignment module (TAM) to improve cross-view consistency before fusion. FAM reduces the discrepancy between infrastructure and vehicle object queries to improve matching reliability, while TAM compensates delayed infrastructure observations through lightweight motion-based temporal alignment. Experiments on V2X-Sim and DAIR-V2X demonstrate consistent improvements over representative baselines under both ideal and delayed settings. Compared with the vehicle-only DQTrack baseline, it improves bicycle and motorcycle AP by 3.1 and 4.8 points. On the real-world DAIR-V2X benchmark, it further improves cyclist detection AP and tracking AMOTA by 2.7 and 3.4 points. These results indicate that aligning heterogeneous and asynchronous V2I information before fusion can improve both benchmark accuracy and the practical reliability of cooperative perception in safety-critical intelligent transportation scenarios.]]></description>
      <pubDate>Mon, 24 Aug 2026 08:46:11 GMT</pubDate>
      <guid>https://trid.trb.org/View/2767157</guid>
    </item>
    <item>
      <title>A Multi-Channel Architecture for Concurrent Safety and Non-Safety V2i Communications</title>
      <link>https://trid.trb.org/View/2732146</link>
      <description><![CDATA[Vehicles-to-Everything (V2X) communications enable real-time data exchange between vehicles and their surroundings, with the aim of improving traffic flow and enhancing road safety. Within this framework, Vehicles-to-Infrastructure (V2I) communication specifically facilitates interactions between vehicles and roadside infrastructure components. Given the diversity of communications present in V2I ecosystems and their differing data quality (DQ) requirements, the simultaneous dissemination of safety and non-safety V2I messages with these varying DQ requirements poses a significant challenge. The IEEE 1609.4 MAC offers a multi-channel operation to address this challenge. However, due to the limited 100ms time slot shared equally between the service channel (SCH) and control channel (CCH), supporting high-frequency safety awareness messages on CCH while managing large packet size non-safety messages results in performance degradation in a shared IEEE 802.11p radio environment. In this study, we present a multi-channel application-aware framework that extends traditional architectures by decoupling periodic safety-critical communications via direct access to CCH and data-intensive non-safety-critical communications powered by IoT application layer protocol (ALPs). Furthermore, alternative transport protocols (QUIC and SCTP) were evaluated as underlying transport for IoT ALPs against Legacy TCP and UDP. Latency, packet delivery ratio (PDR), throughput, inter-arrival time, and connection establishment time were measured in simulation studies. The results showed that ultra-low latency is achieved on the CCH, while QUIC transport-powered IoT ALPs efficiently manage data-intensive payload on SCH.]]></description>
      <pubDate>Wed, 19 Aug 2026 09:27:19 GMT</pubDate>
      <guid>https://trid.trb.org/View/2732146</guid>
    </item>
    <item>
      <title>Adaptive Bat-Inspired Optimization for Infrastructure-Agnostic Task Allocation in Vehicular Fogs</title>
      <link>https://trid.trb.org/View/2732143</link>
      <description><![CDATA[Advancements in intelligent transport systems and cloud computing services have enabled numerous opportunities for sharing computing resources and data to enhance services. Optimizing task allocation in fogs is a major challenge in efficiently utilizing the limited nearby resources. The high-mobility dynamics of urban environments heavily impact the availability of vehicles, resources, and tasks. Several works have explored the allocation problem. However, they focused on specific mobility patterns, failing to consider both the V2I and V2V scenarios, and thus did not observe a more significant aggregation of resources among vehicles. Thus, this work proposes the mechanism SIGHT, which follows the Bat meta-heuristic technique to make task allocation decisions in fogs. Our approach is designed to accommodate V2V and V2I resource allocation, enabling the adaptation of allocation decisions to highly dynamic urban computing scenarios. Extensive simulated analyses have demonstrated the efficiency of SIGHT when compared with optimization techniques, where our approach showed better performance on both infrastructure-less and infrastructure-based comparisons using DBScan and K-means clustering approaches. Results showed that SIGHT allocated up to 9% more tasks, 7 times more resources in a specific scenario (V2I, K-means), and has approximately 8% less failed allocations than other literature algorithms.]]></description>
      <pubDate>Wed, 19 Aug 2026 09:27:19 GMT</pubDate>
      <guid>https://trid.trb.org/View/2732143</guid>
    </item>
    <item>
      <title>A Lightweight Privacy-Preserving Multi-Dimensional Data Aggregation Scheme for the Internet of Vehicles</title>
      <link>https://trid.trb.org/View/2761601</link>
      <description><![CDATA[The Internet of Vehicles (IoV) utilizes intelligent sensing devices installed in vehicles to collect road information. Roadside Units (RSUs) then aggregate the sensed data using statistical methods. These aggregated results assist the traffic management platform in making informed decisions. In the IoV, the collected sensing data is often multi-dimensional. Such fine-grained, multi-dimensional data can more accurately and comprehensively reflect vehicle status. However, this data contains sensitive information such as driving trajectories. Existing privacy-preserving multi-dimensional data aggregation schemes rely on homomorphic encryption and bilinear pairings, which incur high computational costs. Moreover, they exhibit limitations in fine-grained data analysis, resistance to collusion attacks, and fault tolerance. To address these issues, we propose a lightweight privacy-preserving multi-dimensional data aggregation scheme. The scheme leverages the homomorphic properties of secret sharing to achieve efficient multi-dimensional data aggregation and avoids the computational overhead of homomorphic encryption. It employs super-increasing sequences to support fine-grained aggregation across different dimensional subsets. By using batch verification based on small exponents, the verification process for multiple signatures is combined into a single operation, thereby reducing computational costs compared to sequential verification methods. Security analysis and experimental results demonstrate that the proposed scheme effectively protects user privacy, ensures data integrity, and offers lower computational complexity and communication overhead, making it more suitable for highly resource-constrained IoV environments.]]></description>
      <pubDate>Tue, 18 Aug 2026 14:11:27 GMT</pubDate>
      <guid>https://trid.trb.org/View/2761601</guid>
    </item>
    <item>
      <title>Enhancing Autonomous Vehicle Localization Through Cooperative LiDAR and Smart Infrastructure Integration</title>
      <link>https://trid.trb.org/View/2685761</link>
      <description><![CDATA[Autonomous driving systems rely on precise localization to ensure safety and performance in dynamic environments. However, standalone vehicle localization methods using onboard sensors often fail to deliver sufficient accuracy in adverse environments, such as tunnels or areas with limited distinguishable features. To address these challenges, we propose a novel cooperative localization framework that leverages roadside LiDAR-equipped infrastructure and vehicle-to-infrastructure (V2I) communication. In our approach, roadside LiDARs detect and estimate vehicle positions using a refined L-shape fitting algorithm, complemented by the vehicle’s geometric data shared over the V2I network. This method mitigates errors associated with partial point cloud data and improves localization precision significantly. By integrating this infrastructure-based positioning data with onboard localization modules via sensor fusion, our framework enhances overall accuracy and robustness in real-time autonomous driving scenarios. Experimental results in a digital twin environment demonstrate that our system achieves over 70% improvement in localization accuracy compared to self-localization methods. Our work highlights the potential of smart infrastructure to enhance localization, reduce onboard sensor dependency, and improve autonomous driving reliability under challenging environment for self-localization.]]></description>
      <pubDate>Mon, 17 Aug 2026 08:27:43 GMT</pubDate>
      <guid>https://trid.trb.org/View/2685761</guid>
    </item>
    <item>
      <title>Research on Mandatory Lane Changing for Exiting of Buses in V2I Environments: A Centralized Cooperative Control Framework</title>
      <link>https://trid.trb.org/View/2685757</link>
      <description><![CDATA[Bus merging maneuvres after departing stops significantly disrupt urban traffic flow, often causing interruptions and congestion. In order to solve this problem, this paper focuses on the harbour-shaped bus stop scenario and proposes a centralized control framework leveraging V2I coordination technology. This framework comprehensively incorporates multiple complex constraints during vehicle operation, enabling the accurate simulation of real-road driving behaviours. An integrated solving algorithm combining hp-adaptive pseudospectral methods with interior point methods generates optimal control trajectories for vehicles in the target section. Simulation results demonstrate that, compared to the benchmark scheme, the proposed centralized control optimized scheme significantly enhances the timeliness and comfort of bus lane-changing manoeuvres during exit from bus stops, evidenced by reduced lane-changing duration, increased lane-changing speed, and lowered trajectory curvature. Meanwhile, the scheme strictly guarantees the safety of the lane-changing process. Furthermore, the optimization scheme effectively mitigates the interference caused by bus exit manoeuvres on the traffic flow of the road section, leading to a cascading improvement in the travel efficiency of surrounding vehicles. This manifests as shortened average travel time and increased average travel speed for surrounding vehicles. This proves that the method proposed in this paper is beneficial to the overall traffic operation while optimizing the bus operation.]]></description>
      <pubDate>Mon, 17 Aug 2026 08:27:42 GMT</pubDate>
      <guid>https://trid.trb.org/View/2685757</guid>
    </item>
    <item>
      <title>Uncertainty-aware risk assessment and GRU-based risk level map generation for RSU-assisted vehicles</title>
      <link>https://trid.trb.org/View/2728313</link>
      <description><![CDATA[Detecting the surrounding environment and accurately evaluating risks to avoid collisions remain significant challenges in autonomous driving. Recent advances in artificial intelligence (AI) have enhanced trajectory prediction capabilities. However, existing risk assessment methods have not fully integrated the uncertainty inherent in multimodal prediction outputs. Furthermore, relying solely on onboard vehicle devices is challenging, whereas Roadside Units (RSUs), with their prior information, extensive sensor networks, and computing power, provide a more comprehensive risk evaluation. Thus, this paper proposes a cooperative vehicle-infrastructure risk assessment system (CVIRAS). First, an uncertainty-aware risk assessment method based on multimodal prediction results is introduced. By assuming a bivariate Gaussian uncertainty distribution, an approximate collision probability estimation approach is developed, generating risk indicators (RIs) and corresponding risk maps (RMs). Each risk map is then classified to generate a corresponding risk level map (RLM). Utilizing a Gated Recurrent Unit (GRU), an RLM recognition network model (RLMNet) is proposed. This model incorporates a pattern-matching attention mechanism, where predefined patterns with trainable parameters are established. An attention mechanism enables self-learning of the match between each preset pattern and the ego vehicle’s specific state. Furthermore, the authors propose an end-to-end extension of RLMNet (E2E-RLMNet), which tightly couples trajectory prediction and risk assessment within a unified framework for joint training. The proposed framework is validated on real-world NGSIM and CitySim datasets. Experimental results demonstrate that RLMNet and E2E-RLMNet outperform all compared models in both three-class and five-class risk classification, while effectively mitigating the computational burden induced by multimodal and environmental complexity.]]></description>
      <pubDate>Thu, 13 Aug 2026 17:01:44 GMT</pubDate>
      <guid>https://trid.trb.org/View/2728313</guid>
    </item>
  </channel>
</rss>