{
 "schema": "efe-task-dag/1",
 "taxonomy": "v2",
 "taskKey": "robopaint/acrylic_plumpark",
 "suite": "robopaint",
 "base": "robopaint-acrylic-plumpark-i00",
 "title_zh": "丙烯临摹梅园",
 "summary_zh": "主纸上的整幅梅园临摹画在规定平移配准后满足色差、结构、边缘与覆盖相似度，并满足最多384笔、累计路径及笔触涂鸦占比规则。",
 "embodiment_zh": "两种模式均为固定基座的单臂七自由度Franka Panda，作用臂末端为安装工具：圆头丙烯笔刷固定在手端，无可操作夹爪；笔毛长15毫米，桌上有双排十个颜料井、主纸和练习纸。基座不移动；特权模式直接控制仿真，标准模式通过机器人服务控制同一具身。",
 "task_class": "compositional",
 "analysis_level": "step",
 "spec": {
  "version": "1.2",
  "task": "robopaint-acrylic-plumpark-i00",
  "bench": "robopaint",
  "progress": "live",
  "entities": {},
  "milestones": [
   {
    "id": "brush_loaded",
    "zh": "笔刷蘸上颜料",
    "mode": "event",
    "when": null,
    "after": [],
    "hold_s": 0,
    "goal": false,
    "freeze_held": false,
    "criterion_zh": "机械臂末端的圆笔刷进入任一颜料井并蘸上颜料：笔尖到井中心的水平距离小于12毫米，且世界高度低于0.028米；场景视角可看见笔尖在双排颜料井内下探，离井后笔毛带该井颜色，随后出现的有色画迹也能确认已蘸料。仅悬在井上方、尚未进入颜料面或仍是干笔均未完成；一次蘸料成立后保持事件已达成，不要求固定颜色或重复次数。",
    "step": {
     "cat": "其他",
     "item": "蘸取颜料"
    },
    "contact": [
     {
      "who": "手持工具",
      "how": "其他",
      "plan": 1,
      "note_zh": "候选：蘸取；机械臂上固定安装的圆笔刷笔毛浸入颜料井接触颜料，靠蘸附而非从下方托起或舀取，现有接触条目没有准确对应。"
     }
    ],
    "optional": false,
    "verify_zh": null,
    "sources": [
     "inputs/task/privileged/tests/instance.json:70-84 well surface z=0.028 m and radius=0.012 m",
     "inputs/task/privileged/tests/instance.json:213 brush starts dry",
     "inputs/grader/efe_robopaint-files/rcb_brush.py:432-438 reset loaded=False",
     "inputs/grader/efe_robopaint-files/rcb_brush.py:274-279 well_at geometry",
     "inputs/grader/efe_robopaint-files/rcb_brush.py:875-884 dip loads paint",
     "inputs/oracle/replay.mp4:0.0-0.3 first painted marks confirm a prior dip"
    ],
    "conf": 6,
    "note_zh": "依据S2、D11保留干笔初态强制的蘸料前置步骤；此处蘸取不是画的一段连续运动，蘸好后可停留再画。类型候选“蘸取颜料”：材料·舀通常表示舀出物质，不能准确表示笔毛蘸附；关系·放入只描述笔尖进井而不保证载料，故用其他。笔刷已经固定在手端，不建立取笔节点；后续换色不另计步骤。"
   },
   {
    "id": "painting_copied",
    "zh": "临摹目标画",
    "mode": "state",
    "when": null,
    "after": [
     "brush_loaded"
    ],
    "hold_s": 0,
    "goal": true,
    "freeze_held": false,
    "criterion_zh": "录像右下的无遮挡主纸图应呈现目标图上部赭红色横向色带、中部米灰色弯曲色块以及下部深灰绿色斜向枝干，颜色、轮廓和覆盖范围均与目标吻合；按主纸整图在两轴各不超过5毫米的整数像素平移配准（最优值落在搜索边界则不平移）后，双方以标准差1毫米模糊，在任一图有颜料区域向外扩2毫米内的平均CIELAB色差≤2.0、90分位色差≤6.0，亮度SSIM≥0.92（窗口标准差1.5毫米）、亮度梯度幅值相关≥0.8、颜料覆盖交并比≥0.95（覆盖值≥0.5视为有颜料），主纸有色接触笔数≤384。明显空白、缺色、错位或上述任一图像条件不满足均未完成，后续覆盖使条件失效则失去本状态；整项成功另外要求主纸累计有色接触路径≤42300毫米，以及终局涂鸦路径占比≤0.1（单笔路径长除以不小于10毫米的包围盒对角线，所得比值>2.5的笔触路径占总路径的比例），不要求提笔、离手、回位或保持时长。",
    "step": {
     "cat": "材料",
     "item": "画/写"
    },
    "contact": [
     {
      "who": "手持工具",
      "how": "压/擦",
      "plan": 1,
      "note_zh": "机械臂端固定圆笔刷的带色笔毛压在主纸上扫出画迹，改变压深、运动方向与速度形成宽窄不同的笔触。"
     }
    ],
    "optional": false,
    "verify_zh": null,
    "sources": [
     "inputs/task/privileged/instruction.md:3-5 copy target with at most 384 strokes",
     "inputs/grader/efe_robopaint-pictures-files/acrylic-plumpark-i00/target.png",
     "inputs/task/privileged/tests/instance.json:218-230 full thresholds",
     "inputs/grader/efe_robopaint-files/rcb_brush_score.py:365-394 registration and scoring overlap",
     "inputs/grader/efe_robopaint-files/rcb_brush_score.py:398-446 image metrics and inked contact count",
     "inputs/grader/efe_robopaint-files/rcb_brush_score.py:454-458 image acceptance",
     "inputs/grader/efe_robopaint-files/rcb_brush_strict.py:171-193 final process rule and conjunction",
     "inputs/oracle/replay.mp4:0.0,5.0,20.0,40.0,59.0 progress and final unobstructed canvas",
     "inputs/oracle/run.json:reward image and process metrics all pass"
    ],
    "conf": 6,
    "note_zh": "整幅图按汇总相似度验收，目标不是指定逐个完成的笔触清单；384是笔数上限，不是必须画384笔。颜色区域、换色和预试练习纸均不拆节点。此状态表示图像及笔数验收；过程方式的附加条件属于全局成功，因无日志只能在判据中明示，不冒充可计算when。末帧笔刷仍在纸上也可通过；不冻结持有状态。"
   }
  ],
  "constraints": [
   {
    "id": "paint_path_exceeded",
    "zh": "绘画路径超限",
    "never": null,
    "criterion_zh": "主纸上所有带颜料且压纸的笔尖路径累计超过42300毫米即违反；按每个物理步采样，路径只增不减，练习纸和空中路径不计。评分器不会立即停机，但终局即使画面吻合也因路径超限否决成功。",
    "consequence": "历史否决",
    "sources": [
     "inputs/task/privileged/tests/instance.json:227 max_path_mm=42300",
     "inputs/grader/efe_robopaint-files/rcb_brush_strict.py:74-91 main-sheet inked paths",
     "inputs/grader/efe_robopaint-files/rcb_brush_strict.py:157-160 path budget",
     "inputs/grader/efe_robopaint-files/rcb_brush_strict.py:185-198 process veto"
    ]
   }
  ],
  "success": null,
  "atoms": [
   {
    "atom": "brush_loaded",
    "level": "new_code",
    "source": "inputs/grader/efe_robopaint-files/rcb_brush.py:437,875-884; log Session.painter.loaded and color per frame as dyn.brush_loaded and dyn.brush_color"
   },
   {
    "atom": "painting_matches_target",
    "level": "new_code",
    "source": "inputs/grader/efe_robopaint-files/rcb_brush_score.py:365-458; log Session.score() fields de_mean, de_p90, ssim_L, edge_corr, paint_iou, n_strokes and registration per frame"
   },
   {
    "atom": "paint_path_mm",
    "level": "new_code",
    "source": "inputs/grader/efe_robopaint-files/rcb_brush_strict.py:74-91,145-160; log process_metrics.path_mm per frame"
   },
   {
    "atom": "scribble_share",
    "level": "new_code",
    "source": "inputs/grader/efe_robopaint-files/rcb_brush_strict.py:171-180; log process_metrics.scribble_share and scribble_ok, including current stroke"
   }
  ]
 },
 "native_metrics": [
  {
   "name": "de_mean",
   "unit": "DeltaE_ab",
   "threshold": 2.0,
   "direction": "lower_better"
  },
  {
   "name": "de_p90",
   "unit": "DeltaE_ab",
   "threshold": 6.0,
   "direction": "lower_better"
  },
  {
   "name": "ssim_L",
   "unit": "ratio",
   "threshold": 0.92,
   "direction": "higher_better"
  },
  {
   "name": "edge_corr",
   "unit": "correlation",
   "threshold": 0.8,
   "direction": "higher_better"
  },
  {
   "name": "paint_iou",
   "unit": "ratio",
   "threshold": 0.95,
   "direction": "higher_better"
  },
  {
   "name": "n_strokes",
   "unit": "count",
   "threshold": 384,
   "direction": "lower_better"
  },
  {
   "name": "path_mm",
   "unit": "mm",
   "threshold": 42300.0,
   "direction": "lower_better"
  },
  {
   "name": "scribble_share",
   "unit": "ratio",
   "threshold": 0.1,
   "direction": "lower_better"
  },
  {
   "name": "de_mean_coarse",
   "unit": "DeltaE_ab",
   "threshold": null,
   "direction": "lower_better"
  },
  {
   "name": "ssim_L_fine",
   "unit": "ratio",
   "threshold": null,
   "direction": "higher_better"
  },
  {
   "name": "region_cov_min",
   "unit": "ratio",
   "threshold": null,
   "direction": "higher_better"
  },
  {
   "name": "region_cov_mean",
   "unit": "ratio",
   "threshold": null,
   "direction": "higher_better"
  },
  {
   "name": "final_reward",
   "unit": "ratio",
   "threshold": null,
   "direction": "higher_better"
  }
 ],
 "open_questions_zh": [
  "inputs/states/index.json的logs为空：所有when、约束never及success均为null（P8）；现有源码中的指标与载料状态不能假装已经记入逐帧日志。视频可判断构图和蘸料，但精确色差、笔数及过程量须结合该次评分器记录；当前无法完成状态日志上的一致性验证。",
  "按S2、D11及CASES.md OPEN-21保留“笔刷蘸上颜料”前置步骤，并将整幅临摹作为一个汇总目标，按P17分类为compositional；不要求逐井、逐色或逐笔完成。蘸料类型仍未定，使用其他·蘸取颜料，接触使用手持工具×其他（候选蘸取），不将蘸附强行等同于舀取。",
  "任务设计问题：评分器除终局图像外还限制绘画方式，图像完全吻合仍可能因路径或涂鸦占比失败；这些限制已在任务文字明确说明（privileged/instruction.md:41-57），按D9保留为全局成功要求及路径约束，不另建“合规绘画”节点。",
  "终局完整成功规则为图像验收五项及n_strokes≤384，同时path_mm≤42300、scribble_share≤0.1（rcb_brush_strict.py:185-193）。scribble_share可因后续正常笔触稀释而下降，评分器没有历史锁存或立即失败，所以只作为终局success的要求，不写成任何时刻超过0.1即历史否决的never约束；success因无日志为null，其判据已在目标节点明示。",
  "模式差异：特权评分器要求两次独立重放均成功、末态哈希一致且实例重建匹配（privileged/grade.py:104-129）；标准模式直接按服务终局画面评分并要求正常终结，无仿真错误（standard/grade.py:29-49；sim/server.py:83-118）。DAG建相同绘画核心，不把重放、提交、实例匹配或基础设施状态当语义节点或未记录的世界状态条件。",
  "未受评的诊断量：4毫米尺度色差、0.5毫米尺度SSIM、逐色区域覆盖最小值与加权均值在本实例没有阈值；final_reward为连续评分，也不单独决定成功。接触倾斜、压深、速度、硬笔柄接触、蘸料次数及交互成本均仅报告，不建额外约束（rcb_brush_score.py:413-465；rcb_brush.py:889-896；privileged/grade.py:71-87）。",
  "成功参考来自特权模式且没有状态日志：run.json的reward报告平均色差1.4999、90分位3.348、SSIM 0.93389、边缘相关0.867965、交并比0.986912、380笔、路径12922.83毫米、涂鸦占比0，录像末帧画面与目标一致；仅用于核对可见结果，不把其色序、380笔或332次蘸料设成必需。"
 ],
 "conf": 6,
 "conf_reason_zh": "文字、干笔初态、两模式共用阈值和成功参考支持两节点及其唯一先后边；但没有逐帧状态日志可绑定或自检精确验收，且蘸料类型存在词表边界，故置信度为6。",
 "review": {
  "status": "agreed",
  "rounds": 1,
  "reviewers": [
   {
    "engine": "claude",
    "model": "claude-sonnet-5-5",
    "round": 1,
    "verdict": "agree",
    "conf": 7,
    "summary_zh": "DAG两节点（蘸料前置+整幅临摹目标）符合任务文字与评分器；无状态日志故when/success为null并已在open_questions说明；路径超限为历史否决约束，scribble_share仅进success判据，符合评分器无锁存的读法。未发现需改内容的问题。",
    "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：['brush_loaded', 'painting_copied']"
  ]
 },
 "provenance": {
  "prompt": "prompts/01_task_dag.md",
  "prompt_sha256": "b8e79c84b73950529a4fa6b988412b2d221d5df4e052adafcd9605afac64aa25",
  "engine": "codex",
  "model": "gpt-6.1-sol",
  "effort": "high",
  "created": "2026-10-29T11:28:05Z",
  "review_prompt": "prompts/02_task_dag_review.md",
  "review_prompt_sha256": "305c2c5feb2340e0c426ebfe883251f74824cbde2c3e230baa9c86fc05c31550",
  "reconcile_prompt": "prompts/05_reconcile.md",
  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
  "calls": [
   {
    "stage": "dag",
    "round": 0,
    "ok": true,
    "space": "work/tasks/robopaint__acrylic_plumpark/01_dag"
   },
   {
    "stage": "dag_review",
    "round": 1,
    "ok": true,
    "space": "work/tasks/robopaint__acrylic_plumpark/02_review_r1"
   }
  ],
  "review_engine": "claude",
  "review_model": "claude-sonnet-5-5",
  "inputs_manifest_sha256": "c2edde90ee3b1bfe27c1a93d7bf76123ce83cd6eaa5746dacd3a55b2bceb2fc5",
  "repo_commit": "80cac7a5a227db930c659c3307a1e79c2a84bc53"
 }
}