cs.DC提交于 2026-10-27已译

复现可变部分：面向域随机化具身仿真的资源共享

Replicate What Varies: Resource Sharing for Domain-Randomized Embodied Simulation

Zixu Hao · Weijun Wang · Xiangyu Li · Liang Mi · Hao Wu · Yunxin Liu · et al.

中文摘要

仿真为具身智能体提供低成本沙箱，使其能够尝试任务、观察结果并积累经验。并发运行多样化仿真环境会消耗大量 CPU 与 GPU 内存，因为独立构建的场景重复保存了跨实例不变的资源。Patchwork 是面向域随机化仿真的资源共享系统，采用 E=T+Δ 表示：共享模板包含公共资源，补丁记录构建目标环境所需的变更。系统将随机化采样与回放分离，并联合选择可复用模板与补丁，以保持各目标环境预设配置。其分层运行时共享编译后的模型结构与不可变资源，同时将可变参数和仿真状态隔离在各实例内。Patchwork 已实现于 robosuite 和 MuJoCo，并在 LIBERO-plus 与 RoboCasa365 工作负载上评测。运行 32 个副本时，相比每个环境独立使用进程，共享模板使 CPU RSS 降低 88–92%。受控实验表明，在支持的共享边界内，内存增长主要取决于模板多样性和独有资源内容，而非实例数量。

关键要点

01独立场景重复保存不变资源，导致并发仿真占用大量内存。
02以 E=T+Δ 表示共享模板与目标环境补丁，分离随机化采样和回放。
03分层运行时共享编译结构与不可变资源，隔离可变参数和仿真状态。
04在 robosuite、MuJoCo 及 LIBERO-plus、RoboCasa365 上完成实现与评测。
0532 个副本下，CPU RSS 相比每环境一进程降低 88–92%。

解读

尚无解读。

原始英文摘要

arXiv:2610.33582v1 Announce Type: cross Abstract: Simulation provides a low-cost sandbox where embodied agents can attempt tasks, observe outcomes, and learn from experience. Running many diverse simulation environments concurrently can consume substantial CPU and GPU memory because independently constructed scenes duplicate resources that remain unchanged across instances. We present Patchwork, a resource-sharing system for domain-randomized simulation, built around the representation $E=T+\Delta$: a shared template contains common resources, and a patch records the mutations needed to construct a target environment. Patchwork separates randomization sampling from replay and selects reusable templates together with patches that preserve each target environment's intended configuration. Its layered runtime shares compiled model structure and immutable assets while keeping mutable parameters and simulation state private to each instance. We implement Patchwork on robosuite and MuJoCo and evaluate it on LIBERO-plus and RoboCasa365 workloads. At 32 replicas, shared-template execution reduces CPU RSS by 88-92% relative to process-per-environment execution. Controlled experiments show how sharing shifts memory growth toward template diversity and unique asset content rather than instance count, within the supported sharing boundaries.