New Approach to Intelligent Pedestrian Detection and Signaling on Crosswalks

Sanguino, Tomás de J. Mateo; Domínguez, José Manuel Lozano; González, Manuel Joaquín Redondo; Martin, Jose Miguel Davila · 2024 · OpenAlex-citations

DOI: 10.1109/tits.2024.3445156

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the critical issue of pedestrian road safety, motivated by rising accident rates in urban areas of Spain and the United States. The authors identify limitations in existing solutions, such as high infrastructure costs, lack of commercial availability, and an inability to distinguish between pedestrians and vehicles, which leads to false positives. To overcome these challenges, the study introduces an intelligent crosswalk system that integrates speed bumps with autonomous light signaling and artificial intelligence (AI) for precise pedestrian detection. The primary objectives are to evaluate the system’s impact on road safety, analyze changes in user behavior (specifically vehicle speed and pedestrian flow), and assess user perceptions of safety. The proposed system consists of modular speed bumps divided into basic and intelligent nodes. Intelligent nodes contain sensors, electronics, and solar panels, operating autonomously without grid connection. The design uses ISO-ORTO polyester resin reinforced with fibers to withstand mechanical stress, achieving an IP68 rating for dust and water resistance. Detection relies on a sensory fusion process combining RADAR and magnetic field sensors. RADAR data undergoes short-time Fourier transform analysis, with extracted features fed into a One-Class Support Vector Machine (SVM) classifier. This unsupervised approach distinguishes pedestrians from other road users by treating non-pedestrian objects as outliers. The system was validated through quantitative comparisons against conventional crosswalks and a prior fuzzy-logic-based system, alongside user opinion surveys. The results demonstrate significant improvements in road safety metrics. The AI-based detection achieved an accuracy rate of 99.11%, validated through Receiver Operating Characteristic (ROC) analysis. Behavioral studies revealed a 46.5% improvement in pedestrian trajectory regularity. Furthermore, the system induced substantial speed reductions: vehicle speeds decreased by 32.83% during the day and 70.6% at night, while pedestrian speeds dropped by 10.24%, indicating more cautious crossing behavior. Users reported a significant improvement in perceived safety and regulatory compliance. The study also notes that industrial manufacturing could reduce the prototype cost from approximately 6,180 EUR to under 2,050 EUR, enhancing scalability. The significance of this work lies in its practical, commercially viable approach to smart infrastructure. By achieving high detection accuracy and substantial behavioral changes without requiring extensive civil works or grid connectivity, the system offers a robust solution for enhancing pedestrian safety. The findings suggest that integrating AI with physical traffic calming measures effectively mitigates risks associated with driver inattention and excessive speed. This approach addresses gaps in current literature regarding real-world implementation and scalability, providing a framework for future intelligent transportation systems that prioritize immediate detection and intervention over predictive modeling alone.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• vru facing ehmi
• pedestrian behavior perception