SafeFlow: Real-Time Text-Driven Humanoid Whole-Body Control via Physics-Guided Rectified Flow and Selective Safety Gating

TL;DR

SafeFlow is a physics-aware, real-time humanoid control framework that generates safe, feasible motions from text prompts by combining physics-guided flow matching with a multi-stage safety gating system.

Contribution

It introduces a novel physics-guided rectified flow model and a three-stage safety gate for robust, real-time humanoid motion generation from text prompts, addressing physical infeasibility and safety issues.

Findings

Outperforms prior diffusion methods in success rate and physical compliance.

Achieves real-time inference with reduced function evaluations.

Maintains diverse and expressive humanoid behaviors from text prompts.

Abstract

Recent advances in real-time interactive text-driven motion generation have enabled humanoids to perform diverse behaviors. However, kinematics-only generators often exhibit physical hallucinations, producing motion trajectories that are physically infeasible to track with a downstream motion tracking controller or unsafe for real-world deployment. These failures often arise from the lack of explicit physics-aware objectives for real-robot execution and become more severe under out-of-distribution (OOD) user inputs. Hence, we propose SafeFlow, a text-driven humanoid whole-body control framework that combines physics-guided motion generation with a 3-Stage Safety Gate driven by explicit risk indicators. SafeFlow adopts a two-level architecture. At the high level, we generate motion trajectories using Physics-Guided Rectified Flow Matching in a VAE latent space to improve real-robot executability, and further accelerate sampling via Reflow to reduce the number of function evaluations (NFE) for real-time control. The 3-Stage Safety Gate enables selective execution by detecting semantic OOD prompts using a Mahalanobis score in text-embedding space, filtering unstable generations via a directional sensitivity discrepancy metric, and enforcing final hard kinematic constraints such as joint and velocity limits before passing the generated trajectory to a low-level motion tracking controller. Extensive experiments on the Unitree G1 demonstrate that SafeFlow outperforms prior diffusion-based methods in success rate, physical compliance, and inference speed, while maintaining diverse expressiveness.