Skip to content

论文arXiv 2611.150209 天前

Genie Sim Benchmark验证Sim-to-Real高保真度:策略评分相关系数达0. ​

本文由 AI 根据公开内容整理,可能有误,请以原文为准:Genie Sim Benchmark: Real-to-Sim Fidelity and Evaluation(arXiv 2611.15020)

核心内容 ​

arXiv论文提出Genie Sim Benchmark,通过22组涵盖16种任务的真实-仿真配对实验及系统辨识研究,量化评估仿真保真度。基于真实数据微调的pi0.5策略在仿真中表现与实机高度一致,描述性评分相关系数达0.9686,MAE为0.0405,RMSE为0.0520,差异区间控制在-0.110至+0.080之间,验证了该基准在任务结果层面的高置信度。

针对G2 crsB机械臂的长时序闭环测试显示,单关节位置RMSE介于0.00614至0.03522 rad,电流推导扭矩RMSE为0.735至4.001 Nm;32.65秒遥操作轨迹经运动控制器指令整形与伺服校准后实现高精度复现。此外,原生PhysX引擎适配成功恢复三个记录交互任务,证明了跨引擎接触迁移与物理响应一致性评估的有效性。

关键看点 ​

  • 策略评估精度:22组任务对中pi0.5策略仿真与实机评分相关系数0.9686,MAE仅0.0405
  • 动力学对齐指标:G2机械臂关节位置RMSE最高0.035 rad,扭矩RMSE上限4.0 Nm
  • 物理引擎适配:原生PhysX修改版成功复现三个交互任务,支持跨引擎接触迁移验证
  • 遥操作复现验证:32.65秒轨迹经指令整形与伺服校准,排除机械辨识干扰评估控制保真度

工程与产业观察 ​

该基准突破了传统仿真仅关注视觉或单一动力学的局限,建立了“任务成功率+物理响应一致性”的双重评估体系。0.9686的策略评分相关性表明Sim-to-Real迁移可信度已达工程可用阈值,为具身智能算法在仿真中的强化训练与零样本部署提供了量化验收标准。

英文摘要 (Abstract) ​

We evaluate Genie Sim Benchmark through paired real-simulation tasks and complementary system-identification studies. Across 22 robot-task pairs spanning 16 task names, task-specific pi0.5 policies fine-tuned on real-robot data achieve a descriptive Score correlation of 0.9686, MAE 0.0405, RMSE 0.0520, with simulation-real differences from -0.110 to +0.080. G2 crsB experiments assess long-horizon closed-loop position and joint-side torque agreement, while five representative object cases examine identification and cross-engine contact transfer. A separate 32.6503 s teleoperation trajectory, excluded from mechanical identification and servo-gain search, evaluates external position targets through simulated motion-controller command shaping and a calibrated position servo. Across both arms, per-joint position RMSE spans 0.00614-0.03522 rad and current-derived torque RMSE spans 0.73545-4.00099 N m. Native PhysX adaptations recover three recorded interaction tasks, with case-dependent alignment gains. These results support task-specific simulation evaluation through joint assessment of task outcomes and physical-response consistency.

作者:Di Yang, Da Huang, Yang Lv, Fan Yang, Yi Lu, Zheyuan Xing, Linting Guo, Jiaen Shen, Wenjun Sun, Maoqing Yao, Qian Wang

基于本页技术方向关联的整机参数、核心零部件、实测报告与同赛道选型横评。

外部来源

本页事实对应的公开页面,以原文为准。(点击将离开本站)

  1. arXiv 2611.15020:Genie Sim Benchmark: Real-to-Sim Fidelity and Evaluation ↗访问 2026-11-14