Gaussian Process Dynamical Models for Human Motion

TLDR

A GPDM is a latent variable model. The authors introduce Gaussian process dynamical models for nonlinear time series analysis to learn human pose and motion models from high‑dimensional motion capture data. The model uses a low‑dimensional latent space with GP‑based dynamics and observation mappings, marginalizes parameters analytically, and is evaluated by comparing four learning algorithms on 50‑dimensional human motion capture data. The GPDM provides a non‑parametric, uncertainty‑aware dynamical model that, even with small datasets, learns an effective representation of nonlinear dynamics in high‑dimensional human motion.

Abstract

We introduce Gaussian process dynamical models (GPDM) for nonlinear time series analysis, with applications to learning models of human pose and motion from high-dimensionalmotion capture data. A GPDM is a latent variable model. It comprises a low-dimensional latent space with associated dynamics, and a map from the latent space to an observation space. We marginalize out the model parameters in closed-form, using Gaussian process priors for both the dynamics and the observation mappings. This results in a non-parametric model for dynamical systems that accounts for uncertainty in the model. We demonstrate the approach, and compare four learning algorithms on human motion capture data in which each pose is 50-dimensional. Despite the use of small data sets, the GPDM learns an effective representation of the nonlinear dynamics in these spaces.