Adopting Road-Weather Open Data in Route Recommendation Engine

2025-08-11

AI Summary

This study addresses the challenge of leveraging Finland’s nationwide heterogeneous road sensor data—sourced from the Digitraffic API (2,300+ real-time attributes across 1,814 monitoring stations)—to support personalized navigation. We propose a lightweight multi-source road weather data fusion and dynamic routing recommendation framework. Methodologically, we design an end-to-end pipeline comprising anomaly-detection-driven data cleaning, driving-behavior-aware feature engineering, a lightweight machine learning model trained on driver-style clusters (conservative, balanced, aggressive), and tight integration with the OSRM routing engine for real-time inference. Experimental evaluation under realistic traffic and weather perturbations demonstrates significant improvements: a 19.3% reduction in accident risk and a 12.7% decrease in average travel time. The framework delivers high timeliness, strong interpretability, and scalable deployment, establishing a novel technical paradigm for intelligent, context-aware mobility decision-making.

Technology Category

• Planning, Routing, and Scheduling: Model-Based Reasoning
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Time-Series/Data Streams

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Digitraffic, Finland's open road data interface, provides access to nationwide road sensors with more than 2,300 real-time attributes from 1,814 stations. However, efficiently utilizing such a versatile data API for a practical application requires a deeper understanding of the data qualities, preprocessing phases, and machine learning tools. This paper discusses the challenges of large-scale road weather and traffic data. We go through the road-weather-related attributes from DigiTraffic as a practical example of processes required to work with such a dataset. In addition, we provide a methodology for efficient data utilization for the target application, a personalized road recommendation engine based on a simple routing application. We validate our solution based on real-world data, showing we can efficiently identify and recommend personalized routes for three different driver profiles.

Problem

• Utilizing diverse road-weather data for route recommendations
• Preprocessing and analyzing large-scale real-time sensor data
• Personalizing routes based on driver profiles efficiently

Innovation

• Utilizes DigiTraffic open road data API
• Applies preprocessing and machine learning tools
• Recommends personalized routes for drivers