Détection de situations d’homme-à-terre à l’aide d’une plateforme inertielle intra-auriculaire

Abstract

Some industrial areas imply a high risk of accidents at work and isolated work situations where workers in difficulty or inability can not receive immediate assistance without the use of an automatic detection and warning device for these situations of danger to the life or health of workers. However, current solutions have flaws, such as excessice level of false alarms, too long response time or poor ergonomics, which reduce confidence in technology and deployment in the industry. The nature of danger situations is innumerable and complex, from discomfort to exposure to toxic substances, while falls are one of the most important causes of occupational injuries. As the literature does not provide a clear definition of the phenomenon, this project proposes a global definition of man down situations which groups these danger situations according to three observable critical states : the worker falls (F), the worker is immobile (I), the worker is down to the ground (D). A detection strategy is established according to the study of the critical states and the logic of the combinatorial states F-I, F-D, and I-D which boils down to the observation of at least two distinct critical states over a certain period of time. The detection of critical states is based on the body movement and orientation tracking data estimated from the fusion of inertial measurements from a 3-axis accelerometer and a 3-axis gyroscope. A public database of over 4000 records of falls and movements inspired by activities of daily living (ADL) is used for the development of detection methodology and states characterization. The results reveal that the simple detection strategy based on the combinatorial states allows a significant reduction of the false alarms rate to 1,1% for the ADL scenarios, an improvement of the detection rate of the man down situations and thus reach a precision of 99%. The NSERC-EERS Industrial Research Chair in In-Ear Technologies proposes a solution that integrates an inertial measurement unit (IMU) with a the digital custom earpiece to address the related issue of hearing protection for workers as well as improve the workers safety. This project also presents a validation methodology of the for the man down detection solution and the implementation of the in-ear device using a physical tests procedure inspired by the normal activities of the workers. The results of the preliminary tests conducted by three young adult volunteers show good overall performance of the detection algorithms, while confirming the effectiveness of the detection strategy to eliminate false alarms.

Date	18 Jul 2018
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Jérémie Voix (Supervisor) & Bruno De Kelper (Co-supervisor)