Real-time Monitoring of Bicycle Behavior in IoT Fall Detection System

Cycling has been a staple method of transportation in the Philippines, with the government implementing changes to make cycling safer for cyclists. Despite such efforts, bicycle-related accidents are still common in the country. This study is developed to differentiate regular cycling and falling in bicycles using a high-accuracy support vector machine classifier trained with regular cycling and simulated falling, and to distinguish the two from each other to create an accident detection system. In total, ninety-six trials were used to create the SVM Model in MATLAB to produce a classifier with 99.05 percent accuracy. The system was simulated to log sensor data in Google Spreadsheet. For trials with regular cycling, sensor data was stored in Google Sheets, but no SMS was sent. However, in trials where the rider simulated falls, the system successfully alerted their closest relative in real time when falling behavior was detected in the data.