A Neural Temporal Model for Human Motion Prediction

TL;DR

This paper introduces a neural temporal model for human motion prediction that excels at long-term trajectory modeling, uses a novel architecture and loss function, and proposes a new evaluation metric, NPSS, validated through user studies.

Contribution

It presents a new two-level neural architecture, a multi-objective loss, and a novel evaluation metric for improved long-term human motion prediction.

Findings

State-of-the-art long-term motion trajectory modeling

Effective multi-objective loss function for training

NPSS correlates well with human judgment of motion quality

Abstract

We propose novel neural temporal models for predicting and synthesizing human motion, achieving state-of-the-art in modeling long-term motion trajectories while being competitive with prior work in short-term prediction and requiring significantly less computation. Key aspects of our proposed system include: 1) a novel, two-level processing architecture that aids in generating planned trajectories, 2) a simple set of easily computable features that integrate derivative information, and 3) a novel multi-objective loss function that helps the model to slowly progress from simple next-step prediction to the harder task of multi-step, closed-loop prediction. Our results demonstrate that these innovations improve the modeling of long-term motion trajectories. Finally, we propose a novel metric, called Normalized Power Spectrum Similarity (NPSS), to evaluate the long-term predictive ability of motion synthesis models, complementing the popular mean-squared error (MSE) measure of Euler joint angles over time. We conduct a user study to determine if the proposed NPSS correlates with human evaluation of long-term motion more strongly than MSE and find that it indeed does. We release code and additional results (visualizations) for this paper at: https://github.com/cr7anand/neural_temporal_models