🚲 Mobility

'Open Doors, Fix Congestion' — Chengdu traffic police's AI signal and drone campaign

Chengdu overtook Shanghai and Beijing in vehicle registrations by mid-2024, and in October 2024 its traffic police (成都市公安局交通管理局) launched a campaign called '开门治堵' (open doors, fix congestion) combining three tools: deep-learning traffic signals that read live video and recalculate green-light timing in real time instead of running fixed schedules; a 'smart brain' system that flags slow-moving vehicles on ring roads and expressways within 30 seconds; and a drone-and-ground-unit network for faster incident response. Results are corridor-by-corridor, not citywide: the West China Hospital area's congestion index reportedly fell from 3.28 to 2.44 (about 26%), a stretch of Dianxin Road fell about 41%, and one AI-signalled road (Machao Road, ~50,000 vehicles/day) reportedly cut travel time in half. By mid-2025 the city says it freed over 21 million square metres of road space and added 116 traffic lanes.

Live 📅 since 2024

🤖 AI-generated summary of the sources below — how this works · 🔎 Last verified 2026-08-02

🚲 Mobility Traffic management Artificial intelligence Sensors & IoT Congestion pricing

📊 Impact

Congestion index down ~26% near West China Hospital (3.28→2.44) and ~41% on part of Dianxin Road; travel time down ~50% on Machao Road (~50,000 vehicles/day) after AI signal deployment; 21M+ m² of road space freed and 116 lanes added by mid-2025 (all self-reported by traffic police).

🎓 Lesson

Chengdu's AI signals are judged corridor by corridor because the police themselves report results that way — a citywide congestion figure would paper over the fact some streets improved 40%+ while most were untouched.

Evidence🏛️ Official

Scale🏙️ City-wide

📖 Terms in this entry: Adaptive traffic signals · Congestion pricing · GNSS tolling

Sources

↗ 163.com (NetEase) — 'Smart brain' governs slow vehicles set for full rollout; Chengdu's 8-month anti-congestion drive shows clear index dropsPrimary detail on the October 2024 campaign launch, the three technology components, the specific congestion-index and travel-time figures, and the road-space/lane numbers.

↗ Sohu — Chengdu's AI traffic light goes live, cutting travel time 50%Corroborates the AI signal system (deep-learning, real-time video, dynamic green-light calculation) and the Machao Road (3,100m, ~50,000 vehicles/day) 50% travel-time reduction.

↗ CGTN — Smart City Solutions: Chengdu uses AI to tackle traffic in China's most car-crowded cityIndependent international press confirmation that Chengdu leads China in vehicle registrations (since June 2024) and that traffic police are using AI-adjusted signal timing based on live camera feeds; general context, no independent numbers.

This overview was generated with AI from the public sources listed above and checked against them before publication. It has not had a complete human fact-check — treat it as a starting point and follow the sources.

Continue exploring

Compare this project

🇨🇳 TOCC — Chengdu's citywide transport data coordination centre · Chengdu Same city: TOCC's 300M-record data feed is the raw data these AI signals draw their real-time timing from. 🇧🇷 CET traffic control centre and São Paulo's observed congestion index · São Paulo Both cities lead with a congestion index as the headline metric — Chengdu's is corridor-by-corridor, not citywide. 🇺🇸 Surtrac AI Traffic Signals · Pittsburgh Contrast: Chengdu's signals run from a central 'smart brain'; Surtrac lets each intersection negotiate locally, no hub. 🇨🇳 Central Urban Area Intelligent Traffic Management Application · Chongqing Peer AI traffic campaign one province over: Chengdu's signals also read live video feeds, and its results are police-reported too.

Explore the wider context

🛰️ Fifty years of congestion pricing 🚦 Traffic management guide 🏙️ Chengdu city profile 🚲 Mobility projects

📎 Cite this project

1001 Smart Cities (2026). “'Open Doors, Fix Congestion' — Chengdu traffic police's AI signal and drone campaign” — Chengdu, China. The Smart City Atlas. https://1001smartcities.org/projects/chengdu-open-doors-fix-congestion-ai-traffic/ (last verified 2026-08-02). Data: CC BY 4.0.

The underlying data is free to reuse with attribution — see the open data page.

🚩 Report an error