A framework of transportation mode detection for people with mobility disability

Abstract

Transportation mode detection (TMD) is an important computational technique that aids human life at the social and individual levels. However, previous studies on TMD were focused on people without mobility disabilities, and research involving people with mobility disability is limited. Therefore, this study aimed to provide a TMD framework for people with mobility disability. We propose a method for data acquisition, and acquired data pertaining to 120 participants including manual and electric wheelchairs for 15,350 min. We analyzed the acquired data to determine the characteristics of each transportation mode, and applied machine learning and deep learning models to TMD. Our results showed that a recurrent neural network, known as long short-term memory, could classify five transportation modes (still, manual wheelchair, electric wheelchair, subway, and car) for people with and without disabilities, with an accuracy of 96.17%. Our results will be beneficial for enhancing the quality of life and enabling the social inclusion of people with mobility disabilities.