An Integrated Roadside Sensing and Communication Framework for Vulnerable Road User Safety at Signalized Intersections

2026-06-08 · Source: cs.CV updates on arXiv.org · Field: Transportation & Mobility — Autonomous Vehicles & Smart Transportation, Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Expert, short

Summary

An integrated roadside sensing and communication framework is proposed to enhance Vulnerable Road User (VRU) safety at signalized intersections, addressing the fact that VRUs account for approximately half of urban traffic deaths. This framework combines LiDAR, radar, RGB, and thermal cameras for perception, uses edge-based prediction and surrogate-safety analytics for computation, integrates V2X and P2X messaging for communication, and enables adaptive signal control for actuation. An empirical study using the R-LiViT dataset, comprising 200 multi-modal sequences and 2,400 annotated RGB-T frames from three German intersections, revealed VRUs constitute 49% of observations. It also showed a 38% pedestrian and 45% vehicle density drop from day to night, with night-time showing higher close-proximity shares, and 83% of pedestrian bounding boxes being small, indicating distance from sensors. These findings advocate for multi-modal, edge-side, and adaptive solutions.

Key takeaway

For traffic engineers and urban planners focused on reducing urban traffic fatalities, you should prioritize deploying multi-modal sensing and edge-based analytics at signalized intersections. This approach, integrating LiDAR, radar, and thermal cameras with V2X/P2X communication, offers superior VRU protection compared to single-sensor systems. Consider implementing adaptive signal control based on real-time near-miss analytics to dynamically enhance safety and reduce collision risks.

Key insights

Multi-modal sensing and edge analytics significantly improve Vulnerable Road User safety at signalized intersections.

Principles

• VRUs are often distant from sensors, requiring multi-modal coverage.
• Night-time traffic shows higher close-proximity events.
• Adaptive, multi-modal systems outperform uniform single-sensor solutions.

Method

The framework layers include perception (LiDAR, radar, RGB, thermal), computation (edge prediction, surrogate-safety analytics), communication (V2X, P2X), and actuation (adaptive signal control).

In practice

• Deploy multi-modal sensor suites at high-risk intersections.
• Integrate edge-based analytics for real-time near-miss detection.
• Implement V2X/P2X for direct road user warnings.

Topics

• Vulnerable Road Users
• Signalized Intersections
• Multi-modal Sensing
• V2X Communication
• Edge Computing
• Traffic Safety
• R-LiViT Dataset

Best for: Research Scientist, AI Scientist, Computer Vision Engineer, AI Engineer

Related on AIssential

• Decoding Pedestrian Crossing Intention from Egocentric Vision via Vision Language Models — Vision Language Models, fine-tuned with egocentric and gaze cues, significantly improve pedestrian crossing intention prediction.
• CooperDrive: Enhancing Driving Decisions Through Cooperative Perception — Cooperative perception enhances autonomous vehicle safety by expanding situational awareness and enabling proactive decision-making.
• CLEAR: Cognition and Latent Evaluation for Adaptive Routing in End-to-End Autonomous Driving — CLEAR achieves efficient, multi-modal autonomous driving by combining single-step generative planning with LLM-driven cognitive reasoning.
• Camera-RFID Fusion for Robust Asset Tracking in Forested Environments — Fusing camera vision and RFID data enables robust, high-accuracy asset tracking in complex environments by overcoming individual sensor limitations.
• X-Band UAV-enabled Integrated Sensing and Communications for Vehicular Networks — Optimal time allocation in X-band UAV-enabled ISaC systems balances sensing accuracy and communication performance for vehicular networks.
• DAST: A VLM-LLM Framework for Cross-Interface Anomaly Detection in O-RAN — Multi-agent VLM-LLM reasoning, grounded in O-RAN domain knowledge, effectively detects complex cross-interface anomalies in disaggregated networks.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CV updates on arXiv.org.