Neural Assistive Impulses: Synthesizing Exaggerated Motions for Physics-based Characters

TL;DR

This paper introduces a neural control framework that enables physics-based characters to perform exaggerated, stylized motions by reformulating external assistance in impulse space, improving stability and agility.

Contribution

It presents Assistive Impulse Neural Control, a novel approach reformulating assistance in impulse space with hybrid neural policies for stylized motion synthesis.

Findings

Enables robust tracking of highly agile, infeasible maneuvers.

Improves training stability for stylized motion synthesis.

Decomposes assistive signals into analytic and learned components.

Abstract

Physics-based character animation has become a fundamental approach for synthesizing realistic, physically plausible motions. While current data-driven deep reinforcement learning (DRL) methods can synthesize complex skills, they struggle to reproduce exaggerated, stylized motions, such as instantaneous dashes or mid-air trajectory changes, which are required in animation but violate standard physical laws. The primary limitation stems from modeling the character as an underactuated floating-base system, in which internal joint torques and momentum conservation strictly govern motion. Direct attempts to enforce such motions via external wrenches often lead to training instability, as velocity discontinuities produce sparse, high-magnitude force spikes that prevent policy convergence. We propose Assistive Impulse Neural Control, a framework that reformulates external assistance in impulse space rather than force space to ensure numerical stability. We decompose the assistive signal into an analytic high-frequency component derived from Inverse Dynamics and a learned low-frequency residual correction, governed by a hybrid neural policy. We demonstrate that our method enables robust tracking of highly agile, dynamically infeasible maneuvers that were previously intractable for physics-based methods.