Less is More: Saving Manpower on Traffic Control While Improving Road Efficiency

HKUST Dean of Engineering Prof. Hong K. LO and his team have developed an award-winning smart traffic control plan to mitigate the notorious congestion in Kwun Tong District, Hong Kong. And they have a broader vision to share in this story.

Imagine your daily commute trapped in a tailback, watching illegally parked cars block lanes and exacerbate the congestion. Would you instinctively call for more traffic officers to issue tickets?

While that might sound sensible on the surface, HKUST Dean of Engineering Prof. Hong K. LO and his team have developed a smart solution that reduces traffic congestion by, counterintuitively, deploying fewer police officers and traffic wardens to conduct patrols. This has been made possible with artificial intelligence-assisted computation of traffic flow data.

Award-winning traffic management

A project that applies their innovation in Kwun Tong District has successfully shortened commuting time while saving police manpower by 80%. Known as the Smart Traffic Management Operation, the initiative recently earned a Gold Award in the General Service category of the Service Quality Award Scheme 2023, organized by the Service Quality Wing of the Hong Kong Police Force.

“When I drive through the Hoi Yuen Road/Kwun Tong Road roundabout, I’d sometimes be stuck for 10 or even 20 minutes,” Prof. Lo shares. This firsthand experience in Kwun Tong, a business district notorious for traffic congestion, inspired him to act.

A leading expert in intelligent transportation systems and smart cities, Prof. Lo has spent over two decades developing Dynamic Intersection Signal Control Optimization (DISCO), an AI-powered software that replicates real-world traffic to devise customized models that address transportation challenges in various environments. He started to explore how this system might help alleviate congestion in Kwun Tong.

With highly precise traffic flow modeling based on data analytics, DISCO has successfully mitigated congestion in Hoi Yuen Road, Kwun Tong District.

Understanding traffic flow

Through discussions with the district’s police, Prof. Lo learned about their existing traffic management strategies. Typically, about six squads of officers and traffic wardens, or approximately 40 people in total, would be deployed at a time to patrol the roads.

“In collaboration with police, we collected data to study the dynamics on the roads scientifically,” he explains. “For example, how long would it take for officers and traffic wardens to finish a round of patrol, warning off all the drivers parking in unauthorized spots? And how long would it take for those cars to return?”

Such data, combined with traffic light signal timing, were incorporated into the DISCO system. Using the VISSIM multimodal traffic simulation model, the team further analyzed 32 scenarios to find the optimal patrol pattern. They discovered that the deployment of multiple squads to patrol concurrently would backfire, because the illegally parked vehicles would all try to move at the same time, resulting in bottlenecks at intersections.

An unprecedented experiment

After a few months of traffic modeling, the research team put forward a bold plan – sending only one squad at a time to patrol along a carefully designed route that factored in intersection locations and waiting time at traffic lights.

“We’ve never tried this before, and couldn’t find anything similar in existing literature,” says Prof. Lo, who also serves as the founding Director of the GREAT Smart Cities Institute at HKUST.

This unprecedented plan proved successful. Under the new enforcement strategy, illegally parked cars now leave in an orderly manner, clearing the driveway efficiently and effectively – just as the computer model predicted.

Prof. Lo points out that the key to DISCO’s success is the highly precise traffic flow modeling based on data analytics. “On all major roads throughout Hong Kong, the Transport Department has installed cameras to monitor real-time traffic conditions. The rise of edge computing has allowed us to convert such raw footage into meaningful traffic flow data much more efficiently than before. Leveraging machine learning, we can now perform more sophisticated analyses and modeling of traffic flow.”

Start local, think global

The Kwun Tong scheme exemplifies the HKUST team’s ambition to upgrade Hong Kong’s transportation with cutting-edge technology. And it is just one of their many endeavors.

Among other projects, they have introduced smart traffic solutions at congested spots across the city, such as Lung Fu Road roundabout in Tuen Mun, and Tung Chung area-wide traffic control – all using DISCO to adjust traffic signals at intersections to improve traffic flow.

“Very often, we find that congestion can be reduced by 20% to 30% after the adjustments,” Prof. Lo summarizes the achievements. “This is very encouraging.”

Yet his vision extends far beyond geographical boundaries. The team is currently in discussions with the transport authorities in Bangkok, exploring possible ways to mitigate the Thai capital’s severe traffic problems. Additionally, they eagerly anticipate opportunities to put their groundbreaking research to use on the Mainland and in Vietnam.

“I am delighted that Hong Kong is open to experimenting with new technology. This is only the first step, and we hope our local projects will inspire other cities in the world. The potential of intelligent transportation is immense,” Prof. Lo concludes optimistically.

Next time you traverse a formerly jammed road, keep an eye out for an intelligent traffic light developed by HKUST.

The research team uses DISCO to adjust traffic signals at Lung Fu Road roundabout in Tuen Mun for improved traffic flow. (Top video: Before traffic signals were installed at intersections; bottom video: After traffic signals were installed)