Collaborative Bike Sensing for Automatic Geographic Enrichment: Geoannotation of road\/terrain type by multimodal bike sensing

Abstract

In this article, we describe a multimodal bike-sensing setup for automatic geoannotation of terrain types using Web-based data enrichment. The proposed classification system is mainly based on the analysis of volunteered geographic information gathered by cyclists. By using participatory accelerometer and global positioning system (GPS) sensor data collected from cyclists' smartphones, which is enriched with data from geographic Web services, the proposed system is able to distinguish between six different terrain types. For the classification of the Web-based enriched sensor data, the system employs a random decision forest (RDF) (which compared favorably for the geoannotation task against different classification algorithms). The accuracy of the novel bike-sensing system is 92% for six-class road/terrain classification and 97% for two-class on-road/off-road classification. Since the evaluation is performed on large-scale data gathered during real bike runs, these real-life accuracies show the feasibility of our novel approach.