Transport Mode Recognition Data Collection Campaign and Model Training

Car use is one of the main contributors to greenhouse emissions and reduces the quality of the urban space in a city. In the context of Sustainable mobility, tracking and categorizing people's mobility habits is useful for policy, both at the government and corporate levels. This work aims to present the methodology and preliminary results of a Recurrent Neural Network (RNN) model for transportation mode classification that collects data from smartphone users, enriched with open-street map information. Data from 17 test subjects were collected, generating comprehensive classified trip data at the region level. The results based on a portion of the dataset not included in the training are promising.