File size: 14,979 Bytes
deadbe4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
{
 "schema": "efe-task-dag/1",
 "taxonomy": "v2",
 "taskKey": "dexhand/envelope_letter_insert",
 "suite": "dexhand",
 "base": "dexhand-envelope-letter-insert-i00",
 "title_zh": "将折叠信件插入信封",
 "summary_zh": "已三折的信件经封口进入信封袋内、外露不超过 4.95 mm,右手与两物体脱离接触且两物体静止连续 50 个控制步,全程未触发立即判负条件。",
 "embodiment_zh": "两模式均为固定在桌边的天机 Marvin 双臂机器人,每臂七关节;仅右臂及其灵巧手执行任务,左臂与左手停放,底座固定。右手可为 Sharpa Wave(22 个手指执行器,默认)或 Wuji Hand 2.2(20 个);每次只使用一种手,任务几何和判据一致,控制频率 50 Hz。",
 "task_class": "atomic",
 "analysis_level": "step",
 "spec": {
  "version": "1.2",
  "task": "dexhand-envelope-letter-insert-i00",
  "bench": "dexhand",
  "progress": "live",
  "entities": {},
  "milestones": [
   {
    "id": "letter_in_envelope",
    "zh": "信件插入信封",
    "mode": "state",
    "when": null,
    "after": [],
    "hold_s": 0,
    "goal": true,
    "criterion_zh": "总览中原先分离的两张浅色矩形变成信件进入信封袋内的重叠布局,近景可见信件只在封口外留一条窄边,右手已离开:须先经封口进入(前两角越过 x=55 mm 的口线至少 10 mm,|y|≤107.6 mm、高度 1.2–4.4 mm,后两角仍在口外),再使最外角外露≤4.95 mm、四角均满足 |y|≤107.6 mm、x≥-51.6 mm、高度 1.2–4.4 mm,最后信件和信封均无手部接触且两者线速度各≤0.01 m/s、角速度各≤0.2 rad/s,连续 50 个控制步(原生计数 released_hold_steps≥50)。信件仅压在信封表面、仍大段露在口外、角部越出侧缝或手仍在推时都未完成;已达标后退出袋口、重新接触或运动超限会失去该状态,录像末段 stage 3/3、outside 约 4.9 mm、hand contacts L0/E0、released hold 50/50 可辅助确认。",
    "step": {
     "cat": "关系",
     "item": "插入"
    },
    "contact": [
     {
      "who": "手指/夹爪",
      "how": "推/扫",
      "plan": 1,
      "note_zh": "右手指腹在信件表面施加摩擦带动其在桌面滑动,随后以右手指尖推信件后缘入信封口;前后接触顺序见节点说明。"
     }
    ],
    "optional": false,
    "verify_zh": null,
    "sources": [
     "inputs/task.json:10",
     "inputs/task/privileged/instruction.md:3",
     "inputs/task/standard/instruction.md:3",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:36-49",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:95-117",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:139-159",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:167-174",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:248-254",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/5661886e48/runtime.py:131-151",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:10",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:25-72",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/eca9b98723/scene.py:111-175",
     "inputs/oracle/replay.mp4 @ 0-13.28 s"
    ],
    "conf": 6,
    "note_zh": "文本只要求一次插入:到达封口和继续推进为中间过程,松手静置为验收,不建抓取、搬运、抬起、撤手或单独松手节点。信封翻盖已经打开;闭合的弹性口唇并不封死 4 mm 内腔,信件可直接送入,故无独立开口唇前置。初始信件与信封分离,此目标不在起点成立。原生阶段 1、2 锁存但插入几何与最终保持须实时成立;hold_s=0 表示不叠加时间窗,完整 50 步要求由视觉判据和待记录的原生保持计数承载。默认持物冻结适用于本节点要求松手静置的验收;接触标注采用可行的典型推送方案,不要求固定手指或固定滑动路线;README 的 Difficulty 段指出信件薄且初始贴桌,故不臆设可直接抓起的替代方案。"
   }
  ],
  "constraints": [
   {
    "id": "objects_below_desk",
    "zh": "物体低于桌面",
    "never": null,
    "criterion_zh": "信件或信封的物体原点高度低于 0.698 m,即桌面 0.700 m 减去 0.002 m 容差;每个 2 ms 物理子步检查,触发后立即终止。",
    "consequence": "立即终止",
    "sources": [
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:177-183",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:12-15",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:66"
    ]
   },
   {
    "id": "objects_dropped",
    "zh": "物体掉落",
    "never": null,
    "criterion_zh": "信件或信封与地板发生有效接触,或与任何物体均无有效接触且其质心竖直速度小于 -0.1 m/s;无缓冲窗口,每个物理子步立即判负。",
    "consequence": "立即终止",
    "sources": [
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:192-201",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:210-222",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:70"
    ]
   },
   {
    "id": "objects_outside_workspace",
    "zh": "物体离开工作区",
    "never": null,
    "criterion_zh": "信件或信封的原点 x 不在 [0.18, 0.78] m、y 不在 [-0.40, 0.40] m,或 z 大于 1.00 m,即离开允许的桌面工作区。",
    "consequence": "立即终止",
    "sources": [
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:65-67",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:184-187",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:15-24",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:67"
    ]
   },
   {
    "id": "letter_through_envelope",
    "zh": "信件穿透信封",
    "never": null,
    "criterion_zh": "信件与信封的有效碰撞接触穿透深度大于 0.7 mm,或信件中面采样点进入信封后壁、前壁、侧缝、底缝或口唇实体内部;中面按边缘内缩 1 mm 的 7×15 个点检测。",
    "consequence": "立即终止",
    "sources": [
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:52-64",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:202-205",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:212-214",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:225-245",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:71-72"
    ]
   },
   {
    "id": "envelope_displaced",
    "zh": "信封移位过大",
    "never": null,
    "criterion_zh": "信封原点相对于初始桌面坐标 (0.48, 0.16) m 的水平位移范数大于 0.05 m。",
    "consequence": "立即终止",
    "sources": [
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:51",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:188-189",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:25-29",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:68"
    ]
   },
   {
    "id": "envelope_tipped",
    "zh": "信封倾斜过大",
    "never": null,
    "criterion_zh": "信封局部向上轴与世界竖直轴夹角大于 0.35 rad。",
    "consequence": "立即终止",
    "sources": [
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:190-191",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/b3c3eacf51/task.json:69"
    ]
   },
   {
    "id": "forbidden_robot_contact",
    "zh": "非右手接触物体",
    "never": null,
    "criterion_zh": "机器人支架、任一手臂或停放的左手与信件或信封发生有效接触;只有安装于右腕的活动手可接触两物体。",
    "consequence": "立即终止",
    "sources": [
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:68-85",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:206-209",
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/5661886e48/runtime.py:53-65",
     "inputs/task/privileged/instruction.md:3"
    ]
   },
   {
    "id": "unstable_physics",
    "zh": "物理状态失稳",
    "never": null,
    "criterion_zh": "出现 BADQPOS、BADQVEL、BADQACC 警告,或任一广义坐标为非有限数;运行在该物理子步立即终止。",
    "consequence": "立即终止",
    "sources": [
     "inputs/grader/efe_dexhand-files/privileged/environment/variants/5661886e48/runtime.py:108-125"
    ]
   }
  ],
  "success": null,
  "atoms": [
   {
    "atom": "letter_entry_history",
    "level": "new_code",
    "source": "inputs/grader/efe_dexhand-files/privileged/environment/variants/5661886e48/runtime.py:165-166 features.stage_completed / features.stage_steps; log the native mouth-entry stage and its control index"
   },
   {
    "atom": "letter_inserted_geometry",
    "level": "new_code",
    "source": "inputs/grader/efe_dexhand-files/privileged/environment/variants/2bb13ebb1b/checker.py:95-132 geometry.inserted, corners_envelope_frame, outside_m; log these world-state features"
   },
   {
    "atom": "released_rest_hold",
    "level": "new_code",
    "source": "inputs/grader/efe_dexhand-files/privileged/environment/variants/5661886e48/runtime.py:139-167 features.released_hold_steps, hand_letter_contacts, hand_envelope_contacts and both objects speed features; log the native consecutive-control counter"
   },
   {
    "atom": "failure_invariants",
    "level": "new_code",
    "source": "inputs/grader/efe_dexhand-files/privileged/environment/variants/5661886e48/runtime.py:108-125 and :169-170 features.failure / features.failure_detail; log native fail-fast invariant checks for all constraints"
   },
   {
    "atom": "episode_success",
    "level": "new_code",
    "source": "inputs/grader/efe_dexhand-files/privileged/environment/variants/5661886e48/runtime.py:168 features.success; log this native per-frame world-state success flag, not the delivery verifier verdict"
   }
  ]
 },
 "native_metrics": [
  {
   "name": "outside_m",
   "unit": "m",
   "threshold": 0.00495,
   "direction": "lower_better"
  },
  {
   "name": "letter_linear_speed_m_s",
   "unit": "m/s",
   "threshold": 0.01,
   "direction": "lower_better"
  },
  {
   "name": "envelope_linear_speed_m_s",
   "unit": "m/s",
   "threshold": 0.01,
   "direction": "lower_better"
  },
  {
   "name": "letter_angular_speed_rad_s",
   "unit": "rad/s",
   "threshold": 0.2,
   "direction": "lower_better"
  },
  {
   "name": "envelope_angular_speed_rad_s",
   "unit": "rad/s",
   "threshold": 0.2,
   "direction": "lower_better"
  }
 ],
 "open_questions_zh": [
  "inputs/states/index.json 的 logs 为空;按 DECISIONS.md P8,唯一节点 when、全部约束 never 与 success 均为 null,不能把评分器中存在的字段当作已记录字段。所需原生阶段、几何、接触、速度、保持计数和失效原因列为 new_code;当前只能按录像与精确判据判断,无法做状态回放一致性验证。",
  "模式差异:物理场景、checker.py、runtime.py、scene.py 与 task.json 在两模式中逐字一致;privileged 由两次新进程回放成功且末态哈希相同判通过(inputs/grader/efe_dexhand-files/privileged/tests/grade.py:107-124),standard 按最后一次实时回合判定(inputs/grader/efe_dexhand-files/standard/tests/grade.py:22-39)。交付有效性与重放确定性不作为世界状态步骤;此 DAG 建模共同物理成功条件。",
  "录像时间经过加速,原生 50 个控制步不能按播放器 1 秒估计;成功录像末段为静止近景,显示原生保持 50/50。微米级穿透与毫米级四角高度不能由总览可靠量测,需原生特征辅助,不能从表面重叠直接判为入袋。"
 ],
 "conf": 6,
 "conf_reason_zh": "一个插入步骤直接来自两模式文本;阶段、四角边界、放手与静止要求、原生 50 步计数及所有立即失效阈值均可定位到共同评分实现,成功录像展示推入和最终通过。缺少状态日志,细小间隙及失效条件只能保留明确视觉判据和待记录字段,无法检验自动绑定。",
 "review": {
  "status": "agreed",
  "rounds": 1,
  "reviewers": [
   {
    "engine": "claude",
    "model": "claude-sonnet-5-5",
    "round": 1,
    "verdict": "agree",
    "conf": 7,
    "summary_zh": "文本只描述一次“把信件送到封口并送入、然后松手”,单节点插入(松手静置并入 when/criterion)符合 S1/S8;目标条件、阈值、立即终止约束均与评分器和说明一致;无状态日志故 when/never/success 为 null 符合 P8,open_questions 已列出。未发现需修改的内容。",
    "n_issues": 0,
    "n_major": 0
   }
  ],
  "open_zh": []
 },
 "evidence_check": {
  "n_state_runs": 0,
  "n_faithful": 0,
  "success_runs_reaching_all": "0/0",
  "spec_success_eq_grader": "0/0",
  "diverged_runs": [],
  "notes_zh": [
   "when=null 的节点不在状态日志上检查:['letter_in_envelope']"
  ]
 },
 "provenance": {
  "prompt": "prompts/01_task_dag.md",
  "prompt_sha256": "1142cdb56758062bbd0ff585ad99f3fefe07a96377b61614dc4f50b2d16600b7",
  "engine": "codex",
  "model": "gpt-6.1-sol",
  "effort": "high",
  "created": "2026-10-29T02:45:12Z",
  "review_prompt": "prompts/02_task_dag_review.md",
  "review_prompt_sha256": "65812ffc3cb33aaeb10dbe28c925ffe5535a356f8132ec431dcf58b59c2a0b85",
  "reconcile_prompt": "prompts/05_reconcile.md",
  "reconcile_prompt_sha256": "b3cd3803cf22f21db996e7ab1145811f823e92c7728c6f21d3b6649cc1ed0e83",
  "calls": [
   {
    "stage": "dag",
    "round": 0,
    "ok": true,
    "space": "work/tasks/dexhand__envelope_letter_insert/01_dag"
   },
   {
    "stage": "dag_review",
    "round": 1,
    "ok": true,
    "space": "work/tasks/dexhand__envelope_letter_insert/02_review_r1"
   }
  ],
  "review_engine": "claude",
  "review_model": "claude-sonnet-5-5",
  "inputs_manifest_sha256": "00ec9d2db2fc6e7b5daab3c7bfab4ff39b722a3967f54364e44416f20cbcad82",
  "repo_commit": "4c6d01b9990e3b78ddb620bbd237932b3696f60e"
 }
}