Human Motion Prediction via Pattern Completion in Latent Representation Space
This addresses motion understanding for robotics or animation, but it appears incremental as it builds on existing autoencoder and pattern completion ideas.
The paper tackles human motion prediction by using pattern completion in a learned latent space, achieving state-of-the-art performance across multiple tasks without customization.
Inspired by ideas in cognitive science, we propose a novel and general approach to solve human motion understanding via pattern completion on a learned latent representation space. Our model outperforms current state-of-the-art methods in human motion prediction across a number of tasks, with no customization. To construct a latent representation for time-series of various lengths, we propose a new and generic autoencoder based on sequence-to-sequence learning. While traditional inference strategies find a correlation between an input and an output, we use pattern completion, which views the input as a partial pattern and to predict the best corresponding complete pattern. Our results demonstrate that this approach has advantages when combined with our autoencoder in solving human motion prediction, motion generation and action classification.