Physics-constrained Attack against Convolution-based Human Motion Prediction

TL;DR

This paper introduces a physics-constrained adversarial attack method that significantly increases errors in convolution-based human motion prediction models, revealing their vulnerability and suggesting ways to improve robustness.

Contribution

The paper proposes a novel attack method with physical constraints that effectively generates worst-case perturbations tailored to human motion data.

Findings

Prediction errors are significantly increased by the attack.

Current models are vulnerable to the proposed adversarial attack.

Physical constraints improve the naturalness of adversarial examples.

Abstract

Human motion prediction has achieved a brilliant performance with the help of convolution-based neural networks. However, currently, there is no work evaluating the potential risk in human motion prediction when facing adversarial attacks. The adversarial attack will encounter problems against human motion prediction in naturalness and data scale. To solve the problems above, we propose a new adversarial attack method that generates the worst-case perturbation by maximizing the human motion predictor's prediction error with physical constraints. Specifically, we introduce a novel adaptable scheme that facilitates the attack to suit the scale of the target pose and two physical constraints to enhance the naturalness of the adversarial example. The evaluating experiments on three datasets show that the prediction errors of all target models are enlarged significantly, which means current convolution-based human motion prediction models are vulnerable to the proposed attack. Based on the experimental results, we provide insights on how to enhance the adversarial robustness of the human motion predictor and how to improve the adversarial attack against human motion prediction.