Data Assimilation Meetings at Reading

Date Meeting type Speakers
9 April 2014 Internal speaker Emma Villeneuve (University of Reading)
Estimation of limb angular kinematics using particle filter
I will present the SPHERE project (Sensor Platform for HEalthcare in a Residential Environment). This is an interdisciplinary research collaboration (IRC) to develop a platform of tools in order to predict and detect falls and strokes, monitor the activities of daily living, analyse eating behaviour and sleep quality, in a home environment. These tools will be proposed in an interdisciplinary approach, using home-environment sensors, video cameras and wearable sensors. This multi-institutional collaboration has been developed from day one with clinicians, social workers and clinical scientists, in order to solve real healthcare problems in a cost-effective way. At Reading, the project focuses on activity recognition using body-worn sensors (inertial and physiological). From the perspective of inertial sensors, this can be achieved using one or several accelerometer based wearable devices commercially available. The purpose is to determine postures, simple actions such as walking, and eventually activities. The analysis of the gait and position of centre of mass is particularly interesting for Parkinson's patients who fall frequently. Then, a temporal study of activities leads to an individualized pattern of behaviours. This allows detecting abnormal behaviours, which can be changes in the level of disease or an emergency. In the case of a stationary person, moving his arm, the upper arm can be modelled by a link with a fixed joint, which is the shoulder, as the forearm can be modelled by a link with a moving joint, which is the elbow. We want to describe the angular displacement, velocity and acceleration of a link. I will present an estimation method for these 3 variables using a particle filter in a Bayesian framework, some first results and our future projects.

Page navigation