Weakly-supervised Learning of Human Dynamics
This addresses the challenge of analyzing movement quality and efficiency in fields like biomechanics or robotics, where direct dynamics data is scarce, though it is an incremental improvement by leveraging existing motion data with novel neural network layers.
The paper tackles the problem of estimating human dynamics (forces and moments) from motion data, which is laborious to record directly, by proposing a weakly-supervised learning framework that uses easily obtainable motion data and achieves state-of-the-art results in regression tasks like ground reaction force and joint torque.
This paper proposes a weakly-supervised learning framework for dynamics estimation from human motion. Although there are many solutions to capture pure human motion readily available, their data is not sufficient to analyze quality and efficiency of movements. Instead, the forces and moments driving human motion (the dynamics) need to be considered. Since recording dynamics is a laborious task that requires expensive sensors and complex, time-consuming optimization, dynamics data sets are small compared to human motion data sets and are rarely made public. The proposed approach takes advantage of easily obtainable motion data which enables weakly-supervised learning on small dynamics sets and weakly-supervised domain transfer. Our method includes novel neural network (NN) layers for forward and inverse dynamics during end-to-end training. On this basis, a cyclic loss between pure motion data can be minimized, i.e. no ground truth forces and moments are required during training. The proposed method achieves state-of-the-art results in terms of ground reaction force, ground reaction moment and joint torque regression and is able to maintain good performance on substantially reduced sets.