Real Time Pavement Classification Using an Embedded Neural Network

Filipe Do Ó Cavalcanti, Waslon T. A. Lopes, Fabricio Braga Soares de Carvalho

DOI: 10.14209/sbrt.2022.1570824347

Evento: XL Simpósio Brasileiro de Telecomunicações e Processamento de Sinais (SBrT2022)

Keywords: Pavement Classification RTOS Accelerometer Embedded systems

Abstract

The identification of the pavement can be an important feature for the automotive market, as it impacts on different characteristics of the vehicle and its navigability. Considering the importance to identify the pavement, this paper describes the design and implementation of a system capable of data acquisition and real time classification of two types of road pavement: asphalt and paving stone. The system is based on a real-time operating system that polls data from a triaxial accelerometer and GPS at a fixed frequency and offloads it to a computer. An Artificial Neural Network is trained in Python with 92% accuracy and the model is exported to the embedded system for real-time classification while driving.

Download