Learning Variations in Human Motion via Mix-and-Match Perturbation
This addresses the need for more realistic and varied motion predictions in applications like animation or robotics, though it is incremental as it builds on existing stochastic methods.
The paper tackled the problem of generating diverse future human motions from observed poses by introducing a stochastic method to combine random noise with pose information, preventing the model from ignoring the noise. The result is a model that produces higher-quality and more diverse pose sequences compared to state-of-the-art techniques.
Human motion prediction is a stochastic process: Given an observed sequence of poses, multiple future motions are plausible. Existing approaches to modeling this stochasticity typically combine a random noise vector with information about the previous poses. This combination, however, is done in a deterministic manner, which gives the network the flexibility to learn to ignore the random noise. In this paper, we introduce an approach to stochastically combine the root of variations with previous pose information, which forces the model to take the noise into account. We exploit this idea for motion prediction by incorporating it into a recurrent encoder-decoder network with a conditional variational autoencoder block that learns to exploit the perturbations. Our experiments demonstrate that our model yields high-quality pose sequences that are much more diverse than those from state-of-the-art stochastic motion prediction techniques.