How AI and Big Data Are Transforming Urban Traffic Management

At the recent 2017 12th China Intelligent Transportation Annual Conference, experts discussed city intelligent transport, autonomous driving, vehicle‑road collaboration, rail transit intelligence, road safety technologies, traffic big data, smart highways, water transport, traffic organization optimization, and the "AI + traffic" innovation.

The conference keywords were "system thinking, AI, innovation," indicating a trend toward top‑level design, AI‑driven technology applications, and micro‑innovation in traffic organization.

1. Top‑Level Design in Three Steps

First, improve front‑end equipment to enhance perception: Deploy more cameras, sensors, and upgrade existing analog/SD cameras to HD, filling perception blind spots.

Second, data aggregation and unified analysis: Consolidate video and image data from NVRs and other devices to the cloud for centralized scheduling, leveraging cloud computing platforms such as Huawei Video Cloud.

Third, big‑data analysis and AI innovation: Structure video and image data, detect violations, and apply AI‑driven analytics to create new traffic applications, requiring strong AI computation and high‑performance big‑data platforms.

2. AI Creates Traffic Benefits

Traffic vision has evolved through three stages: "visible" (analog/SD cameras), "clear" (HD cameras), and "understandable" (AI‑enhanced cameras). Many cities have completed HD upgrades and now focus on AI applications that can replace manual detection and perform tasks beyond human capability.

AI is used for image recognition to identify violations, handling millions of images that would be impossible to process manually. Reinforcement learning enables modeling of traffic patterns from macro regions to micro intersections, offering powerful new use cases such as autonomous driving, AI‑enabled police vehicles, and AI‑signal interaction for optimal road rights management.

3. Intersection Innovation Improves Efficiency

Intersection control is critical for overall traffic flow. Cities like Shenzhen, Chengdu, and Xi'an have experimented with queue‑type flow, variable lanes, tidal lanes, left‑turn borrowing, zipper alternation, and double‑wait stacking to boost throughput.

Combining AI with traffic simulation allows validation of these innovative intersection strategies, with several cities already seeing promising results.

Conclusion

The conference served as a capstone for 2017, summarizing best practices and forecasting 2018 trends. Continued innovation and practice are expected to make urban traffic safer, more orderly, smoother, and smarter, with AI and big‑data at the core of this transformation.