标题： RoboPearls：机器人操作的可编辑视频仿真 显示英文标题

标题： RoboPearls: Editable Video Simulation for Robot Manipulation

摘要： 通用机器人操作策略的发展取得了显著进展，这得益于在多样化环境中的大规模演示数据。 然而，收集现实世界演示的高成本和低效率阻碍了数据获取的可扩展性。 虽然现有的仿真平台为机器人学习提供了受控环境，但弥合仿真到现实之间的差距仍然是一个挑战。 为了解决这些挑战，我们提出了RoboPearls，这是一个可编辑的视频仿真框架，用于机器人操作。 基于3D高斯点云（3DGS），RoboPearls能够从演示视频中构建逼真的、视角一致的仿真，并支持广泛的仿真操作符，包括各种物体操作，这些操作由先进的模块如增量语义蒸馏（ISD）和3D正则化NNFM损失（3D-NNFM）提供支持。 此外，通过结合大型语言模型（LLMs），RoboPearls通过灵活的命令解释和执行，以用户友好的方式自动化仿真生产过程。 此外，RoboPearls采用视觉-语言模型（VLM）分析机器人学习问题，以闭合仿真循环来提高性能。 为了展示RoboPearls的有效性，我们在多个数据集和场景上进行了广泛的实验，包括RLBench、COLOSSEUM、Ego4D、Open X-Embodiment和一个真实世界的机器人，这些实验展示了我们令人满意的仿真性能。 显示英文摘要

摘要： The development of generalist robot manipulation policies has seen significant progress, driven by large-scale demonstration data across diverse environments. However, the high cost and inefficiency of collecting real-world demonstrations hinder the scalability of data acquisition. While existing simulation platforms enable controlled environments for robotic learning, the challenge of bridging the sim-to-real gap remains. To address these challenges, we propose RoboPearls, an editable video simulation framework for robotic manipulation. Built on 3D Gaussian Splatting (3DGS), RoboPearls enables the construction of photo-realistic, view-consistent simulations from demonstration videos, and supports a wide range of simulation operators, including various object manipulations, powered by advanced modules like Incremental Semantic Distillation (ISD) and 3D regularized NNFM Loss (3D-NNFM). Moreover, by incorporating large language models (LLMs), RoboPearls automates the simulation production process in a user-friendly manner through flexible command interpretation and execution. Furthermore, RoboPearls employs a vision-language model (VLM) to analyze robotic learning issues to close the simulation loop for performance enhancement. To demonstrate the effectiveness of RoboPearls, we conduct extensive experiments on multiple datasets and scenes, including RLBench, COLOSSEUM, Ego4D, Open X-Embodiment, and a real-world robot, which demonstrate our satisfactory simulation performance.