具身智能世界模型评测榜单发布，Seedance 2.0暂居第一|EmbodiedCLUE-World

{  "depth_accuracy": [    0.4658589680989583,    [      {        "video_path": "/worldmodel/WorldArena/video_quality/data_by_model/Seedance2.0/generated_dataset/PnPBottleToCabinetClose/episode_000001/1/video",        "video_results": 0.95458984375,        "video_results_normalized": 0.8235928347154255      },      ...    ]  ],  "trajectory_accuracy": [    9.16235,    [      {        "video_path": "/worldmodel/WorldArena/video_quality/data_by_model/Seedance2.0/generated_dataset/PnPBottleToCabinetClose/episode_000001/1/video",        "video_results": 7.25,        "video_results_normalized": 0.17746120330934548      },      ...    ]  ],  "motion_smoothness": [    1.6298582802216213,    [      {        "video_path": "/worldmodel/WorldArena/video_quality/data_by_model/Seedance2.0/generated_dataset/PnPBottleToCabinetClose/episode_000001/1/video",        "video_results": 1.906134843826294,        "video_results_normalized": 0.7216654086344958      },      ...    ]  ],  "subject_consistency": [    0.006557115598125077,    [      {        "video_path": "/worldmodel/WorldArena/video_quality/data_by_model/Seedance2.0/generated_dataset/PnPBottleToCabinetClose/episode_000001/1/video",        "video_results": 0.884976786095649,        "video_results_normalized": 0.884976786095649      },      ...    ]  ],  "logics": {    "generated_dataset_PnPBottleToCabinetClose_episode_000001_1": false,    "generated_dataset_PnPBottleToCabinetClose_episode_000002_1": false,    ...    "generated_dataset_PnPCupToDrawerClose_episode_000001_1": false,    "generated_dataset_PnPCupToDrawerClose_episode_000002_1": true,    ...   },  "semantic_alignment": [    0.8507975333333334,    [      {        "video_path": "/worldmodel/WorldArena/video_quality/data_by_model/Seedance2.0/generated_dataset/PnPBottleToCabinetClose_episode/000001/1/video",        "video_results": 0.790527,        "video_results_normalized": 0.790527      },      ...    ]  ]}

[  {    "video": "PnPBottleToCabinetClose_episode_000001.mp4",    "metrics": {      "Subgoal_Completion": {        "score": 5,        "reason": "机器人成功完成了接近瓶子、抓取、抬起、搬运至柜子上方、放置、释放以及最后关闭柜门的全部子目标动作。",        "score_normalized": 1.0,        "subgoals": {          "approach_object": true,          "grasp_object": true,          "lift_object": true,          "transport_to_target": true,          "place_into_target": true,          "release_object": true,          "close_container": true        }      },      "Contact_Causality": {        "score": 5,        "reason": "瓶子的运动轨迹与机器人的抓取、移动和放置动作完全吻合，柜门的关闭也是由机械臂的物理推力导致的，物理因果关系清晰合理。",        "score_normalized": 1.0      },      "Failure_Mode_Diagnosis": {        "primary_failure": "No Failure",        "secondary_failures": [],        "severity": "none",        "reason": "视频完整且高质量地展示了任务的执行过程，没有发现明显的物理错误或逻辑失败。"      }    },    "raw_response_file": "/worldmodel/WorldArena/video_quality/tmp_extra_VLM/Seedance2.0/PnPBottleToCabinetClose_episode_000001_20260520_151726.json",    "error": null,    "judge_backend": "gemini_openai"  },  {    "video": "PnPBottleToCabinetClose_episode_000001.mp4",    "metrics": {      "Instruction_Following": {        "score": 5,        "reason": "机器人完整地执行了抓取瓶子、将其放置在柜子隔板上并随后撤离的动作序列，符合指令要求。虽然采样帧中未显示关闭柜门的动作，但搬运和放置的核心任务已清晰完成。",        "score_normalized": 1.0      },      "Interaction_Quality": {        "score": 4,        "reason": "机器人抓取和放置瓶子的过程较为自然，接触点稳定。但在放置过程中，机械臂与柜子边缘存在轻微的视觉重叠或穿模嫌疑，整体交互质量良好。",        "score_normalized": 0.8      },      "Perspectivity": {        "score": 5,        "reason": "视频视角始终保持固定，场景中的柜子、台面和瓶子的空间布局、尺度以及遮挡关系在整个过程中保持高度一致，没有出现视角崩坏或物体漂浮现象。",        "score_normalized": 1.0      }    },    "raw_response_file": "/worldmodel/WorldArena/video_quality/tmp_VLM/Seedance2.0/PnPBottleToCabinetClose_episode_000001_20260520_150819.json",    "error": null,    "judge_backend": "gemini_openai"  },  ...]

"dimension_scores": {      "Task_Completion": {        "name_zh": "任务遵循",        "weight": 0.05,        "metrics": [          "Instruction_Following",          "Subgoal_Completion",          "Final_State_Similarity"        ],        "score": 0.941221      },      "Physical_Interaction": {        "name_zh": "物理交互",        "weight": 0.05,        "metrics": [          "Interaction_Quality",          "Contact_Causality"        ],        "score": 0.963333      },      "Trajectory_Motion": {        "name_zh": "轨迹运动",        "weight": 0.3,        "metrics": [          "Trajectory_Accuracy",          "Motion_Smoothness"        ],        "score": 0.420669      },      "Spatial_Geometry": {        "name_zh": "空间一致",        "weight": 0.05,        "metrics": [          "Perspectivity",          "Depth_Accuracy"        ],        "score": 0.96674      },      "Content_Consistency": {        "name_zh": "内容语义",        "weight": 0.35,        "metrics": [          "Subject_Consistency",          "Semantic_Alignment"        ],        "score": 0.873905      },      "Failure_Diagnosis": {        "name_zh": "结果诊断",        "weight": 0.2,        "metrics": [          "Failure_Mode_Diagnosis"        ],        "score": 0.983333      }    },    "overall_valid_score_normalized": 0.772