Work

Detection of equestrian falls using smartphone sensors

Year: 2017
Type: dissertation
Abstract: Horseback riding is enjoyed by millions of people worldwide, both as a competitive sport and as recreational activity. Horses have a congenital habit to flee from potential danger and the nature of the sport carries with it a risk of falling from the horse. Equestrians are often riding alone and the additional risk of being unaided after a severe fall accident is prominent. The need of a smartphone application able to sense a fall and automatically contact preselected relatives is identified. An equestrian fall detection algorithm using smartphone sensor data is developed in this master thesis. The work was made from scratch, including data recording of both normal horseback riding activities and simulated fall events. Additional signal representations were derived and features calculated. Their contribution of interesting information was evaluated and the most suitable features made the final threshold based fall detection algorithm. The algorithm was implemented as an Android application running on real time sensor data for evaluation. In total, 53 out of 55 simulated falls were detected and no false alarm was obtained during eleven hours of normal riding activity. The algorithm is considered as suitable for its purpose of increasing equestrian safety.
Author: Malin Magnusson
Institution: Lund University
Language: English
Open Access status: green
OpenAlex ID: W2727698407
Cites: 15
Cited by:
Related to: 10