Foundation Intelligence for Smart Infrastructure Services in Transportation 5.0

IEEE Transactions on Intelligent Vehicles, Journal Year: 2024, Volume and Issue: 9(1), P. 39 - 47

Published: Jan. 1, 2024

This perspective paper delves into the concept of foundation intelligence that shapes future smart infrastructure services as transportation sector transitions era Transportation 5.0. First, discussion focuses on a suite emerging technologies essential for intelligence. These encompass digital twinning, parallel intelligence, large vision-language models, traffic simulation and systems modeling, vehicle-to-everything (V2X) connectivity, decentralized/distributed systems. Next, introduces present landscape 5.0 applications illuminated by foundational casts vision towards including cooperative driving automation, intersection/infrastructure, management, virtual drivers, mobility planning operations, laying out prospects are poised to redefine ecosystem. Last, through comprehensive outlook, this aspires offer guiding framework intelligent evolution in data generation model calibration, twinning simulation, scenario development experimentation, feedback loop management control, continuous learning adaptation, fostering safety, efficiency, reliability, sustainability infrastructure.

Language: Английский

---

Towards the Next Level of Vehicle Automation Through Cooperative Driving: A Roadmap From Planning and Control Perspective

IEEE Transactions on Intelligent Vehicles, Journal Year: 2024, Volume and Issue: 9(3), P. 4335 - 4347

Published: Feb. 8, 2024

Cooperative Driving Automation (CDA) stands at the forefront of evolving landscape vehicle automation, elevating driving capabilities within intricate real-world environments. This research aims to navigate path toward future CDA by offering a thorough examination from perspective Planning and Control (PnC). It classifies state-of-the-art literature according classes defined Society Automotive Engineers (SAE). The strengths, weaknesses, requirements PnC for each class are analyzed. analysis helps identify areas that need improvement provides insights into potential directions. further discusses evolution directions CDA, providing valuable enhancement enrichment research. suggested include: robustness against disturbance; Risk-aware planning in mixed environment Connected Automated Vehicles (CAVs) Human-driven (HVs); Vehicle-signal coupled modeling coordination enhancement; Vehicle grouping enhance mobility platooning.

Language: Английский

---

Smart Mobility Digital Twin Based Automated Vehicle Navigation System: A Proof of Concept

IEEE Transactions on Intelligent Vehicles, Journal Year: 2024, Volume and Issue: 9(3), P. 4348 - 4361

Published: Feb. 21, 2024

Digital twins (DTs) have driven major advancements across various industrial domains over the past two decades. With rapid in autonomous driving and vehicle-to-everything (V2X) technologies, integrating DTs into vehicular platforms is anticipated to further revolutionize smart mobility systems. In this paper, a new DT (SMDT) platform proposed for control of connected automated vehicles (CAVs) next-generation wireless networks. particular, enables cloud services leverage abilities promote experience. To enhance traffic efficiency road safety measures, novel navigation system that exploits available information designed. The SMDT are implemented with state-of-the-art products, e.g., CAVs roadside units (RSUs), emerging cellular V2X (C-V2X). addition, proof-of-concept (PoC) experiments conducted validate performance. performance evaluated from standpoints: (i) rewards on and, xmlns:xlink="http://www.w3.org/1999/xlink">(ii) latency reliability platform. Our experimental results using SUMO-based large-scale simulations show can reduce average travel time blocking probability due unexpected incidents. Furthermore, record peak overall modeling route planning be 155.15 ms 810.59 ms, respectively, which validates our design aligns 3GPP requirements use cases fulfills targets design. demonstration video found at https://youtu.be/3waQwlaHQkk .

Language: Английский

---

Multisensor Fusion for Vehicle-to-Vehicle Cooperative Localization With Object Detection and Point Cloud Matching

IEEE Sensors Journal, Journal Year: 2024, Volume and Issue: 24(7), P. 10865 - 10877

Published: Feb. 19, 2024

Accurate vehicle pose is fundamental information required by automated driving systems. However, complicated enironments and sensor failures have constrained onboard sensor-based single-vehicle localization precision. With the development of cooperative automation, from surrounding vehicles in vehicle-to-vehicle (V2V) network offers remarkable potential to boost ego vehicle's performance. In this article, we propose a framework based on multisensor fusion that uses shared multiagents, leveraging point cloud feature matching object detection. The detection system can determine relative between corresponding data LiDAR sensor. accuracy derived directly deep-learning-based limited. Thus, refining method proposed further improve applying technique normal distribution transformation approach. Meanwhile, reduce transmission load, extract only edge plane features scan exclude remaining cloud. Additionally, fused into inertial navigation (INS)-based system, which enables continuous high-frequency output within Kalman filter framework. To make algorithm more adaptive different noise levels, measurement quality evaluation rule designed. Real-world vehicular experiments show at least 35% compared traditional range-based method.

Language: Английский

---

Cooperative Localization in Transportation 5.0

IEEE Transactions on Intelligent Vehicles, Journal Year: 2024, Volume and Issue: 9(3), P. 4259 - 4264

Published: March 1, 2024

In the era of future mobility within Transportation 5.0, autonomy and cooperation across all road users smart infrastructure stand as key features to enhance transportation safety, efficiency, sustainability, supported by cooperative perception, decision-making planning, control. An accurate robust localization system plays a vital role in enabling these modules for is constrained environmental uncertainties sensing limitations. To achieve precise resilient this new era, paper introduces emerging technologies including edge computing, hybrid data-driven physical model approaches, foundation models well parallel intelligence, that are beneficial next-generation systems. On top technologies, integrating real-world testing digital twin technology, we further put forward Decentralized Autonomous Service (DAS)-based framework systems resilience, robustness, safety

Language: Английский

---

V2X-Real: A Largs-Scale Dataset for Vehicle-to-Everything Cooperative Perception

Lecture notes in computer science, Journal Year: 2024, Volume and Issue: unknown, P. 455 - 470

Published: Nov. 28, 2024

Language: Английский

---

Seamless Virtual Reality With Integrated Synchronizer and Synthesizer for Autonomous Driving

IEEE Robotics and Automation Letters, Journal Year: 2024, Volume and Issue: 9(5), P. 4218 - 4225

Published: March 14, 2024

Virtual reality (VR) is a promising data engine for autonomous driving (AD). However, fidelity in this paradigm often degraded by VR inconsistency, which the existing approaches become ineffective, as they ignore inter-dependency between low-level synchronizer designs (i.e., collector) and high-level synthesizer processor). This paper presents seamless virtual ( $\mathsf {SVR}$ ) platform AD, mitigates such enabling agents to interact with each other shared symbiotic world. The crux an integrated {IS}^{2}$ design, consists of drift-aware lidar-inertial colocation motion-aware deep visual synthesis network augmented image generation. We implement on car-like robots two sandbox platforms, achieving cm-level colocalization accuracy 3.2% deviation, thereby avoiding missed collisions or model clippings. Experiments show that proposed reduces intervention times, turns, failure rates compared benchmarks. -trained neural can handle unseen situations real-world environments, leveraging its knowledge learnt from space.

Language: Английский

---

Toward Foundation Models for Inclusive Object Detection: Geometry- and Category-Aware Feature Extraction Across Road User Categories

IEEE Transactions on Systems Man and Cybernetics Systems, Journal Year: 2024, Volume and Issue: 54(11), P. 6570 - 6580

Published: April 24, 2024

The safety of different categories road users comprising motorized vehicles and vulnerable (VRUs) such as pedestrians cyclists is one the priorities automated driving smart infrastructure services. Three-dimensional (3-D) LiDAR-based object detection has been a promising approach to perceiving users. Despite accurate 3-D geometry information, point cloud from LiDAR usually nonuniform, learning effective abstract representations for diverse remains challenging detection, particularly small objects VRUs. For inclusive (IDetect), we propose general foundation convolution component, called geometry-aware (GA Conv) toward feature extraction model, serve basic operations neutral network detection. Further, GA Conv are then utilized elementary layers build novel elegant pyramid IDetect. It learns geometric-related features unstructured data by implicitly distribution property geometry-related in particular proposed IDetect comprehensively evaluated on large-scale benchmark Waymo open datasets with all qualitative quantitative experiment results demonstrate that can effectively consider nonuniform distributed clouds learn geometric assist user In addition, integrated other state-of-the-art neural networks performance boost VRU demonstrated, showing functionality making it component future model.

Language: Английский

---

HPL-ViT: A Unified Perception Framework for Heterogeneous Parallel LiDARs in V2V

Published: May 13, 2024

Language: Английский

---

Lateral Velocity Estimation Utilizing Transfer Learning Characteristics by a Hybrid Data-mechanism-driven Model

2022 IEEE Intelligent Vehicles Symposium (IV), Journal Year: 2024, Volume and Issue: unknown, P. 460 - 465

Published: June 2, 2024

Language: Английский

---