Surfaces categorization based on data collected by bike sensors

Abstract

Today computing is being applied in several areas of knowledge and, when it is used with dynamic technologies as Internet of Things and Artificial Intelligence, can take users experience to a higher level. This work, for example, proposes an application in the context of smart cities to analyze data intelligently, knowing that, the concept of Smart Cities involves a wide range of innovations created for the comfort of the citizens. This paper proposes, through data collection, the recognition of vibratory patterns for the classification of surfaces using machine learning techniques. This is an important issue, as it offers a proposal for greater security of bicycle circulation points with the identification of possible irregularities. The analysis of roads surface quality is possible with the use of an accelerometer to collect data important for the audition of tracks. This data is then classified generating information, classified as patterns (asphalt and pavement surfaces). We have performed field data gathering and applied algorithms calculations to classify data to identify the surface the bicycle ridden, with results in percentages of accuracy up to more than 96%.