# FARM: Frame-Accelerated Augmentation and Residual Mixture-of-Experts for Physics-Based High-Dynamic Humanoid Control

## TL;DR

FARM is a comprehensive framework that improves physics-based humanoid control during high-dynamic actions by combining frame-accelerated data augmentation, a robust base controller, and a residual mixture-of-experts, setting new benchmarks in the field.

## Contribution

The paper introduces FARM, a novel end-to-end framework that significantly enhances high-dynamic humanoid motion tracking and provides the first open benchmark for this challenge.

## Key findings

- FARM reduces tracking failure rate by 42.8%.
- FARM lowers mean per-joint position error by 14.6%.
- FARM maintains high accuracy on low-dynamic motions.

## Abstract

Unified physics-based humanoid controllers are pivotal for robotics and character animation, yet models that excel on gentle, everyday motions still stumble on explosive actions, hampering real-world deployment. We bridge this gap with FARM (Frame-Accelerated Augmentation and Residual Mixture-of-Experts), an end-to-end framework composed of frame-accelerated augmentation, a robust base controller, and a residual mixture-of-experts (MoE). Frame-accelerated augmentation exposes the model to high-velocity pose changes by widening inter-frame gaps. The base controller reliably tracks everyday low-dynamic motions, while the residual MoE adaptively allocates additional network capacity to handle challenging high-dynamic actions, significantly enhancing tracking accuracy. In the absence of a public benchmark, we curate the High-Dynamic Humanoid Motion (HDHM) dataset, comprising 3593 physically plausible clips. On HDHM, FARM reduces the tracking failure rate by 42.8\% and lowers global mean per-joint position error by 14.6\% relative to the baseline, while preserving near-perfect accuracy on low-dynamic motions. These results establish FARM as a new baseline for high-dynamic humanoid control and introduce the first open benchmark dedicated to this challenge. The code and dataset will be released at https://github.com/Colin-Jing/FARM.