FLASH: Fast Learning via GPU-Accelerated Simulation for High-Fidelity Deformable Manipulation in Minutes

TL;DR

FLASH is a GPU-optimized deformable object simulation framework enabling high-fidelity, real-time manipulation learning, which significantly accelerates policy training and transfer to real robots.

Contribution

The paper introduces FLASH, a novel GPU-native simulation engine designed specifically for contact-rich deformable manipulation, achieving high accuracy and speed on large-scale soft-body simulations.

Findings

FLASH scales to over 3 million degrees of freedom at 30 FPS on a single GPU.

Policies trained solely in FLASH transfer zero-shot to real robots performing complex tasks.

FLASH enables rapid policy training in minutes, reducing reliance on real-world data.

Abstract

Simulation frameworks such as Isaac Sim have enabled scalable robot learning for locomotion and rigid-body manipulation; however, contact-rich simulation remains a major bottleneck for deformable object manipulation. The continuously changing geometry of soft materials, together with large numbers of vertices and contact constraints, makes it difficult to achieve high accuracy, speed, and stability required for large-scale interactive learning. We present FLASH, a GPU-native simulation framework for contact-rich deformable manipulation, built on an accurate NCP-based solver that enforces strict contact and deformation constraints while being explicitly designed for fine-grained GPU parallelism. Rather than porting conventional single-instruction-multiple-data (SIMD) solvers to GPUs, FLASH redesigns the physics engine from the ground up to leverage modern GPU architectures, including optimized collision handling and memory layouts. As a result, FLASH scales to over 3 million degrees of freedom at 30 FPS on a single RTX 5090, while accurately simulating physical interactions. Policies trained solely on FLASH-generated synthetic data in minutes achieve robust zero-shot sim-to-real transfer, which we validate on physical robots performing challenging deformable manipulation tasks such as towel folding and garment folding, without any real-world demonstration, providing a practical alternative to labor-intensive real-world data collection.