Transportation mode identification and real-time CO2 emission estimation using smartphones: How CO2GO works

Abstract

In a context where personal mobility accounts for about two thirds of the total transportation energy use, assessing an individual’s personal contribu- tion to the emissions of a city becomes highly valuable. Prior efforts in this direction have resulted in web-based CO2 emissions calculators, smartphone- based applications, and wearable sensors that detect a user’s transportation modes. Yet, high energy consumption and had-hoc sensors have limited the potential adoption of these methodologies. In this technical report we outline an approach that could make it possible to assess the individual carbon footprint of an unlimited number of people. Our application can be run on standard smartphones for long periods of time and can operate transparently. Given that we make use of an existing platform (smartphones) that is widely adopted, our method has the potential of unprecedented data collection of mobility patterns. Our method estimates in real-time the CO2 emissions using inertial in- formation gathered from mobile phone sensors. In particular, an algorithm automatically classifies the user’s transportation mode into eight classes us- ing a decision tree. The algorithm is trained on features computed from the Fast Fourier Transform (FFT) coefficients of the total acceleration measured by the mobile phone accelerometer. A working smartphone application for the Android platform has been developed and experimental data have been used to train and validate the proposed method.