Dual Kalman filter for estimating load-free human motion kinematic energy expenditure

The need for measuring energy expenditure using non-wearable devices in sports science is a complex task involving strict protocols of measurement. Such protocols and measurement involving indirect calorimetry or inertial measurement unit (IMU) based measurement are expensive to setup, too inaccurate or force the subjects being measured to modify their actions in a significant manner. In this paper, we explored the concept of using a parallel Kalman filter setup being used in simulation. The simulation used was an iterative one using a dynamic model, featuring the use of two different types of setup- structured or non structured movement and the use of one or two Kalman filters. The results from the simulations performed showed that structured movement using a dual Kalman filter setup was the best performer when using root mean square error as the metric for performance. These results will help influence our work utilising the Microsoft Kinect and estimating weights of human joints and the energy expenditure attained from that.