Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →China standard: Multi Pillar. # "多支柱"标准综合应用指南
.claude/skills/pangzhenying2025-china-multi-pillar/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 69% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 30% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -9% | 0% |
| 属性 | 值 | |------|-----| | 名称 | 智能网联汽车自动驾驶系统"多支柱"标准综合应用指南 | | 状态 | 工作组征求意见稿（2024年10月） | | 推荐等级 | P2-推荐入选 | | 总页数 | 152页 | | 特殊价值 | 中国独有系统观，多标准协同实施方法论 |
"多支柱"方法发展历程
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
2019年1月: WP.29/GRVA提出"三支柱"测试方法
├── 支柱1: 实际道路测试
├── 支柱2: 封闭场地测试
└── 支柱3: 审核评估（含仿真）
发展为"多支柱":
├── 支柱1: 场景目录 (Scenario Catalogue)
├── 支柱2: 模拟仿真 (Simulation)
├── 支柱3: 封闭场地 (Proving Ground)
├── 支柱4: 实际道路 (Real World Test)
├── 支柱5: 审核评估 (Audit & Assessment)
└── 支柱6: 在用监测 (In-Service Monitoring)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━调研的自动驾驶产品
├── 城市干支路产品
│ ├── 4家乘用车企业 + 1家商用车企业
│ ├── 最高设计车速: 50-70 km/h
│ └── 功能: 跟车/避撞/换道/转弯/掉头/环岛/隧道
├── 高快速路产品
│ ├── 13家企业（乘用车92.3% + 牵引车7.7%）
│ └── ODD: 高速/城市快速路/匝道/桥梁/隧道
└── 当前技术限制
├── 天气: 横风/大雪/积水识别不足
├── 道路: 破损路面识别不足
└── 信号: 潮汐车道/可变车道/待转区仍在开发多支柱指南涉及的国家标准
├── GB/T 40429-2021 汽车驾驶自动化分级
├── GB/T 44721-2024 自动驾驶系统通用技术要求
├── GB/T（待定）自动驾驶系统设计运行条件
├── GB/T（待定）自动驾驶功能仿真试验方法及要求
├── GB/T 41798-2022 场地试验方法及要求
└── GB/T 44719-2024 道路试验方法及要求多支柱综合应用框架 (第四章)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
各支柱的实际应用能力分析
├── 仿真试验：高覆盖率，但模型保真度有限
├── 场地试验：高控制性，但场景有限
├── 道路试验：高真实性，但覆盖率低
├── 审核评估：系统性评价，但主观性较强
└── 在用监测：持续性数据，但滞后性
综合应用原则:
1. 各支柱互补，不可单一替代
2. 仿真为主体（覆盖面），实车为验证（关键场景）
3. 审核评估贯穿全过程
4. 在用监测实现闭环
基于综合应用框架的现有标准使用:
→ 附录A提供了详细的应用规程（94页起）
→ 包含具体操作指南和案例
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━skills/china-standards/ads-safety/ — ADS安全要求skills/china-standards/scenario-safety/ — 场景安全评估skills/china-standards/odd/ — ODD标准skills/automotive-scenario-driven-testing/ — 场景驱动测试方法| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 22,100 | 22,259 | +1% | 1 | 1 | 0% | 3,831 | 4,859 | +27% | 0 | 0 | — |
case-10 | pass→pass | 15,831 | 16,322 | +3% | 1 | 1 | 0% | 2,378 | 3,717 | +56% | 0 | 0 | — |
case-11 | fail→pass | 6,757 | 5,657 | -16% | 1 | 1 | 0% | 1,154 | 1,949 | +69% | 0 | 0 | — |
case-02 | fail→pass | 22,486 | 13,729 | -39% | 1 | 1 | 0% | 3,876 | 3,653 | -6% | 0 | 0 | — |
case-03 | fail→pass | 27,123 | 27,083 | -0% | 1 | 1 | 0% | 4,364 | 5,680 | +30% | 0 | 0 | — |
case-04 | fail→pass | 16,795 | 10,604 | -37% | 1 | 1 | 0% | 2,693 | 2,452 | -9% | 0 | 0 | — |
case-05 | fail→pass | 21,309 | 5,937 | -72% | 1 | 1 | 0% | 3,764 | 2,117 | -44% | 0 | 0 | — |
case-12 | fail→pass | 31,694 | 7,030 | -78% | 1 | 1 | 0% | 6,173 | 2,273 | -63% | 0 | 0 | — |
case-06 | pass→pass | 20,332 | 14,053 | -31% | 1 | 1 | 0% | 3,063 | 3,140 | +3% | 0 | 0 | — |
case-07 | fail→fail | 15,826 | 14,629 | -8% | 1 | 1 | 0% | 2,517 | 3,368 | +34% | 0 | 0 | — |
case-08 | fail→pass | 19,380 | 19,339 | -0% | 1 | 1 | 0% | 2,984 | 4,113 | +38% | 0 | 0 | — |
case-09 | fail→fail | 19,434 | 21,225 | +9% | 1 | 1 | 0% | 2,989 | 4,112 | +38% | 0 | 0 | — |
case-13 | fail→pass | 19,083 | 14,706 | -23% | 1 | 1 | 0% | 3,071 | 3,603 | +17% | 0 | 0 | — |
case-14 | fail→pass | 22,754 | 13,790 | -39% | 1 | 1 | 0% | 3,229 | 3,235 | +0% | 0 | 0 | — |
case-15 | fail→pass | 17,889 | 8,130 | -55% | 1 | 1 | 0% | 2,682 | 2,248 | -16% | 0 | 0 | — |
case-16 | fail→pass | 13,400 | 3,778 | -72% | 1 | 1 | 0% | 2,234 | 1,648 | -26% | 0 | 0 | — |
case-17 | fail→pass | 13,658 | 5,933 | -57% | 1 | 1 | 0% | 2,283 | 1,974 | -14% | 0 | 0 | — |
case-18 | fail→pass | 7,716 | 1,981 | -74% | 1 | 1 | 0% | 1,379 | 1,313 | -5% | 0 | 0 | — |
case-19 | pass→pass | 16,294 | 3,538 | -78% | 1 | 1 | 0% | 2,399 | 1,496 | -38% | 0 | 0 | — |
case-20 | pass→pass | 19,981 | 45,065 | +126% | 1 | 1 | 0% | 3,539 | 4,952 | +40% | 0 | 0 | — |
case-21 | pass→pass | 24,216 | 26,877 | +11% | 1 | 1 | 0% | 4,146 | 5,488 | +32% | 0 | 0 | — |
case-22 | pass→pass | 17,805 | 23,434 | +32% | 1 | 1 | 0% | 3,005 | 4,819 | +60% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +64 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.