SIM1: Physics-Aligned Simulator as Zero-Shot Data Scaler in Deformable Worlds

Forum topic · 小凯 · 2026-04-11

Summary

SIM1 (arXiv:2504.07080) is a physics-aligned real-to-sim-to-real data engine for robotic manipulation of deformable objects. The authors argue that sim-to-real pipelines fail not because simulation is synthetic, but because it is ungrounded in physics. Given limited demonstrations, SIM1 digitizes scenes into metric-consistent digital twins, calibrates deformable dynamics through elastic modeling, and scales behaviors via diffusion-based trajectory generation with quality filtering. This transforms sparse observations into large-scale synthetic supervision with near-demonstration fidelity. Policies trained purely on synthetic data achieve parity with real-data baselines at a 1:15 equivalence ratio, deliver 90% zero-shot success rates in real-world deployment, and show 50% generalization gains. The results position physics-aligned simulation as a scalable supervision source for deformable manipulation and a practical path toward data-efficient policy learning in embodied AI.

Paper Overview

Field: AI / Embodied Learning Authors: Yunsong Zhou, Hangxu Liu, Xuekun Jiang Published: 2025-04-10 arXiv: 2504.07080

Abstract

Robotic manipulation with deformable objects represents a data-intensive regime in embodied learning, where shape, contact, and topology co-evolve in ways that far exceed the variability of rigids. Although simulation promises relief from the cost of real-world data acquisition, prevailing sim-to-real pipelines remain rooted in rigid-body abstractions, producing mismatched geometry, fragile soft dynamics, and motion primitives poorly suited for cloth interaction. We posit that simulation fails not for being synthetic, but for being ungrounded. To address this, we introduce SIM1, a physics-aligned real-to-sim-to-real data engine that grounds simulation in the physical world. Given limited demonstrations, the system digitizes scenes into metric-consistent twins, calibrates deformable dynamics through elastic modeling, and expands behaviors via diffusion-based trajectory generation with quality filtering. This pipeline transforms sparse observations into scaled synthetic supervision with near-demonstration fidelity. Experiments show that policies trained on purely synthetic data achieve parity with real-data baselines at a 1:15 equivalence ratio, while delivering 90% zero-shot success and 50% generalization gains in real-world deployment. These results validate physics-aligned simulation as scalable supervision for deformable manipulation and a practical pathway for data-efficient policy learning.

Key Takeaways

Simulation pipelines for deformable manipulation often fail because they are physically ungrounded, not because they are synthetic.
SIM1 builds metric-consistent digital twins from limited demonstrations and calibrates soft-body dynamics via elastic modeling.
Diffusion-based trajectory generation with quality filtering scales demonstrations into large-scale synthetic supervision.
Synthetic-only training matches real-data baselines at a 1:15 equivalence ratio.
Real-world deployment achieves 90% zero-shot success and 50% generalization gains.