量身 AI← 案例库← Case library
案例 · AI 落地方案CASE · AI DEPLOYMENT PLAN · 城市公共交通 · 多伦多Urban transit · Toronto

多伦多公车局 TTCToronto Transit Commission

从试点线路走向全网均匀运行From pilot routes to a network that runs evenly
基于公开资料的分析,与该企业没有合作关系。Based on public information; no affiliation with the company.
视频制作中,上线后会出现在这里。Video coming soon — it will appear here.
一句话判断The verdict 试点分散在路线优化、轨道安全与绕行;要规模化,需要一个由运营部门负责的实测结果。Pilots span routing, track safety and detours; scaling needs one measured result that operations owns.
起步顺序Order of work 先在试点线路测扎堆,与对照线路比较,赢得调度与司机认可,再扩大。Measure bunching on pilot routes, compare with control routes, win operators over, then widen.

公开事实Public facts 已核对Checked

  • 419.8M
    2024 年载客量Riders carried in 2024较 2023 年增长约 6%Up about 6% on 2023
  • 6%
    较 2023 年的客流增长Ridership growth over 2023准点与人力压力Pressure on punctuality and crews
  • 700+
    AI 每月标记的绕行数Detours flagged each month by AI与 Transit App 的一年期试点One-year pilot with Transit App
  • 5
    2026 年 3 月启动的研究项目Research projects launched in March 2026与 TMU、DMZ 合办的 Transit Innovation YardTransit Innovation Yard, with TMU and DMZ

从试点线路到全网的阶梯A ladder from pilot route to network 判断Judgment

  • 试点线路Pilot routes:对照基线测量扎堆与断档。Measure bunching and gapping against a baseline.
  • 对照线路Control routes:与不用 AI 的相似线路比较。Compare with similar routes that do not use AI.
  • 一线接受度Operator acceptance:调度员与司机试用并商定规则。Dispatchers and operators test it and agree the rules.
  • 全网The network:经安全与劳资审查后才扩大。Widen only after safety and labour review.

专家组裁决Expert panel 依据公开观点的推演Inference from public views

Foteini Agrafioti

技术与业务战略对齐Strategy alignment

每个试点挂到一个运营目标上。Tie each pilot to one operating goal.

Erik Brynjolfsson

价值与度量Value & measurement

先定基线和对照线路,再算收益。Baseline and control routes before counting gains.

Ziad Obermeyer

算法在优化什么What it optimizes

轨道入侵预警:按地点与时段检查漏报。Intrusion alerts: check misses by location and time.

Ethan Mollick

人怎么用How people use it

先让调度员在真实扰动里试。Let dispatchers test on real disruptions first.

Andrew Ng

节奏与方法Pace & method

一组线路,三到六个月出数字。One route group, numbers in three to six months.

Thomas Davenport

整家企业怎么变Whole-enterprise change

改的是调度流程,不是工具;点名负责人。This changes dispatch, not tools. Name the owner.

保留的分歧Disagreement kept · 速度 vs. 安全审查Speed vs. safety review

Ethan Mollick:尽早上真实扰动场景。Get tools onto real disruptions early.

Ziad Obermeyer:预警先过目标与误差关。Alerts need target and error review first.

→ 绕行走快速通道;轨道预警走完整安全审查。Fast lane for detours; full safety review for track alerts.

六位专家是谁、怎样思考、有什么铁律 →Who the six experts are, how they think, their rules →

三个目标Three priorities 建议Recommendation

1

让一个结果有主人Make one result owned

在试点线路上测扎堆与断档,运营部门负责,并有基线。Bunching and gapping on pilot routes, with an operations owner and a baseline.

2

用对照线路证明Prove it against control routes

扩大前与不用 AI 的线路比较。Compare with routes that do not use AI before widening.

3

赢得一线接受Win operator acceptance

调度员与司机参与制定 AI 建议的使用规则。Dispatchers and operators help write the rules for AI advice.

场景组合Scenarios 建议Recommendation

场景Scenario做法Approach状态Status
扎堆与断档管理Bunching and gapping control
提效Efficiency体验Experience
向调度员建议保持与间距;由调度员决定。先在试点线路。Suggest holds and spacing to dispatchers; they decide. Start on pilot routes.建议 · 董事会已要求探索Proposed · board asked to explore
动态路线优化Dynamic route optimisation
提效Efficiency
测试结合路况、中断与客流的研究原型。Test the research prototype that blends traffic, disruptions and demand.研究原型(公开)Research prototype (public)
轨道入侵预警Track intrusion alerts
降风险Risk
摄像头加 AI 提示轨道有人;须与安全治理部门确认。Camera and AI alert for people on the track; confirm with safety governance.试点(公开)· 安全审查Pilot (public) · safety review
预测性维护Predictive maintenance
提效Efficiency降风险Risk
用车辆与轨道状态数据,在故障前安排维修。Use fleet and track condition data to plan repairs before failures.建议Proposed

12 个月路线12-month roadmap 建议Recommendation

奠基Foundation

第 1–3 月Months 1–3
  • 试点线路基线Baselines on pilot routes
  • 选定对照线路Control routes picked
  • 点名运营负责人Operations owner named
◆ 基线定稿Baseline final

试点Pilot

第 4–6 月Months 4–6
  • 试点线路调度建议Dispatch advice on pilot routes
  • 司机与调度员试用反馈Operators test and comment
  • 首份实测报告First measured report
◆ 指标达成Targets met

上线Go live

第 7–9 月Months 7–9
  • 与一线商定规则Rules agreed with operators
  • 更多线路,同口径More routes, same measures
  • 上线即监控Monitoring from day one
◆ 安全签字Safety sign-off

扩展Extend

第 10–12 月Months 10–12
  • 全网就绪评审Network readiness review
  • 维护案例初稿Maintenance case drafted
  • 年度重排优先级Next year re-ranked
◆ 进入下一年Next year begins

治理与合规Governance 建议Recommendation

  • 轨道预警:扩大前须与安全治理部门确认Track alerts: confirm with safety governance before any widening
  • 劳资与用工影响:向法律顾问确认Labour and workforce implications: confirm with counsel
  • 由调度员决定,AI 只提建议Dispatchers decide; AI advises

三层指标Three-layer metrics 建议Recommendation

采用Adoption每班使用建议的调度员Dispatchers using advice per shift · 试点覆盖的线路Routes covered by the pilot
过程Process被调度员采纳的建议Advice accepted by dispatchers · 试点线路的间隔波动Headway variance on pilot routes · 预警误报率Alert false-alarm rate
结果Outcome与对照线路相比的扎堆情况Bunching versus control routes · 乘客等候时间Wait time for riders

目标值由企业自定。Targets are the company's to set.

风险Risks 判断Judgment

收益来自别处Gains that came from elsewhere

时刻表调整可能被误当成 AI 的功劳;对照线路可以区分。Timetable changes can look like AI wins; control routes tell them apart.

一线不信任建议Operators distrust advice

若建议像监控就会被无视;尽早让他们参与。If advice feels like monitoring, it will be ignored; involve them early.

预警漏报Alerts that miss

漏报比误报更严重;两类都要审查。A missed track alert matters more than a false one; review both kinds.

前 90 天First 90 days 建议Recommendation

第 1–2 周Weeks 1–2点名运营负责人;盘点试点Name the operations owner; list pilots
第 3–4 周Weeks 3–4收集试点线路基线Collect baselines on pilot routes
第 5–10 周Weeks 5–10选对照线路;起草一线规则Pick control routes; draft operator rules
第 13 周Week 13首次复盘First review
今天只需要一个决定:One decision today: 指定一组线路及其运营负责人name one route group and its operations owner

我们可能错在:We could be wrong if: 调度员在真实扰动中否定建议,或对照线路不用 AI 也同样改善。dispatchers reject the advice on real disruptions, or control routes improve just as much without AI.

本页是基于公开资料的分析,与所述企业没有合作关系,不代表其官方立场或内部计划;企业数字多来自其自述或媒体转述;「专家组」是依据六位专家公开观点的推演,不是他们的原话,也不代表他们或其雇主的立场。资料整理日期:2026-11-14。This page is an analysis based on public information. It has no affiliation with the company and does not represent its official position or internal plans. Company figures come mostly from its own statements or media reports. The expert panel is an inference from six experts' public views, not their words, and does not represent them or their employers. Compiled 2026-11-14.

想给你的公司也做一份?Want one for your company?

同样的方法:公开事实、专属判断、12 个月路线和前 90 天。公司大小都可以聊,中文沟通。Same method: public facts, a company-specific judgment, a 12-month roadmap and the first 90 days. Any size of business; we can talk in Chinese or English.

Telegram @johnsunlei邮件 Email [email protected]