Human Motion Reconstruction using Wearable Accelerometers

We address the problem of capturing human motion in scenarios where the use of a traditional optical motion capture system is impractical. Such scenarios are relatively commonplace, such as in large spaces, outdoors or at competitive sporting events, where the limitations of such systems are apparent: the small physical area where motion capture can be done and the lack of robustness to lighting changes and occlusions. In this paper, we advocate the use of body-worn wearable wireless accelerometers for reconstructing human motion and to this end we outline a system that is more portable than traditional optical motion capture systems, whilst producing naturalistic motion. Additionally, if information on the person's root position is available, an extended version of our algorithm can use this information to correct positional drift.