Dhaka rolls out AI traffic enforcement to curb notorious congestion

Dhaka, the capital of Bangladesh, is notorious for its dense population and severe traffic congestion. The city has recently welcomed its first AI (Artificial Intelligence) traffic enforcement system, hoping to improve years of poor driving order.

According to AFP, buses, cars, motorcycles, and rickshaws in Dhaka often compete for road space, treating traffic signals, pedestrian crossings, and even traffic police as obstacles.

On-the-spot fines often lead to arguments, and conflicts are common. Sometimes, officers are even rammed by vehicles refusing to obey commands.

Traffic police officer SM Nazim Uddin said, "Those who violate the rules will argue back with us," "But since AI was introduced, drivers have started obeying the regulations, and we are spared the daily arguments."

Despite multiple attempts to modernize traffic management, the city of over 22 million people has always relied on manual control, with police officers even using ropes to stop vehicles before traffic light changes.

However, since April, Dhaka police have been using artificial intelligence software combined with traffic surveillance cameras to automatically detect various violations.

Hannan Rahman Jibon, a 28-year-old driver, was one of the first drivers caught. He said, "I ran a red light, and the car owner, sitting at home, would receive a text message saying this car violated the rules." Jibon was automatically issued a fine of 2,000 taka (Bangladeshi currency), equivalent to NT$506.

"Now I am more careful because cameras have been installed in many places."

A study by the US National Bureau of Economic Research listed Dhaka as the "slowest city" in the world.

According to research by the World Bank and the Bangladesh University of Engineering and Technology, the average speed in Dhaka is only 4.8 kilometers per hour, which is slower than walking.

N.M. Nasiruddin, spokesperson for the city police, said the AI system uses existing traffic surveillance camera footage, and the software can identify various violations such as running red lights, illegal lane changes, and illegal parking.

The system's coverage does not yet extend to the entire city, but manual enforcement will be gradually phased out. Police say the system has already begun to change driver behavior.

"We have started to see results," Nasiruddin said.

Inside the control room at police headquarters, 52-year-old analyst Sharmin Afroze monitors a row of real-time traffic conditions, which are connected to the AI system, and the system then flags violations for manual review.

Afroze said, "Previously, police had to stop cars, check documents, and determine fines." Now, police issue fines to the most serious offenders and warnings to others.

Despite this, officials acknowledge that the technology faces challenges.

"We are also encountering some difficulties," Afroze said. "For example, some license plates are blurry or too small to be identified."

Police are cooperating with road transport authorities to address this issue, and other functions, such as detecting vehicles driving on sidewalks, will be added successively.