PulseAugur
EN
LIVE 15:42:04

New diffusion models enhance humanoid control with natural language

By PulseAugur Editorial · [3 sources] ·

Two new research papers introduce advanced diffusion models for controlling physics-based humanoids using natural language. SCRIPT utilizes a multi-stage training framework with a Joint Action-State-Text Diffusion Transformer and Reinforcement Learning with Hybrid Rewards to improve instruction following and motion quality. MIND employs a multi-scale intent diffusion mechanism, using behavioral intent as a bridge between text commands and low-level actions, and demonstrates improved semantic alignment and physical plausibility. AI

IMPACT These models represent advancements in embodied AI, potentially enabling more sophisticated and intuitive control of robots through natural language instructions.

RANK_REASON Two academic papers published on arXiv introduce new methods for controlling physics-based humanoids.

Read on arXiv cs.LG →

AI-generated summary · Google Gemini · from 3 sources. How we write summaries →

New diffusion models enhance humanoid control with natural language

COVERAGE [3]

  1. arXiv cs.LG TIER_1 English(EN) · Jingyan Zhang, Han Liang, Ruichi Zhang, Bin Li, Juze Zhang, Xin Chen, Jingya Wang, Lan Xu, Jingyi Yu ·

    SCRIPT: Scalable Diffusion Policy with Multi-stage Training for Language-driven Physics-Based Humanoid Control

    arXiv:2605.22894v1 Announce Type: cross Abstract: Controlling physics-based humanoids from natural-language instructions is a critical step toward general-purpose embodied agents. However, existing methods remain constrained by a tension between semantic expressiveness and physic…

  2. arXiv cs.CV TIER_1 English(EN) · Bin Li, Ruichi Zhang, Han Liang, Jingyan Zhang, Juze Zhang, Xin Chen, Jingya Wang ·

    MIND: Multi-Scale Intent Diffusion for Text-Driven Physics-Based Humanoid Control

    arXiv:2605.26006v1 Announce Type: new Abstract: Enabling physics-based humanoids to execute diverse behaviors from high-level textual commands remains a significant challenge. Existing methods typically follow either a two-stage paradigm that combines kinematic motion generation …

  3. arXiv cs.CV TIER_1 English(EN) · Jingya Wang ·

    MIND: Multi-Scale Intent Diffusion for Text-Driven Physics-Based Humanoid Control

    Enabling physics-based humanoids to execute diverse behaviors from high-level textual commands remains a significant challenge. Existing methods typically follow either a two-stage paradigm that combines kinematic motion generation with physics-based tracking, or an end-to-end im…

RELATED ENTITIES

RELATED TOPICS