An AI-based traffic management system was installed at the Gulshan-2 intersection in the capital in June 2023 with the aim of controlling traffic and enforcing traffic laws using technology. But, the pilot project, which cost Tk60.47 lakh, has not been effectively operational due to a lack of coordination among the relevant authorities in the absence of a specific “policy framework”.

According to officials concerned, the Dhaka North City Corporation (DNCC) initiated the system, with only DNCC officials having access to the AI camera-generated videos. However, they lack the authority to enforce traffic laws.

On the other hand, the traffic division of police – responsible for managing the traffic system – has not been granted access to the videos or control over signal lights, leading to chaos at the intersection.
Speaking to the Daily Sun, a traffic police sergeant at Gulshan-2 intersection, wishing to remain anonymous, revealed that no cases have ever been filed based on the AI camera footage.

AI eyes see all, but can’t enforce“We manually file cases against traffic law violators since we lack access to the AI camera-produced videos. Traffic management is also conducted manually due to drivers’ non-compliance with traffic signals, VIP movements, and other issues requiring manual intervention,” he explained.

The sergeant further commented on the inefficiencies of the system, stating, “Although the system supposedly counts vehicles and signals automatically, we’ve observed severe traffic congestion extending from Gulshan-2 to Banani, yet the red signals persist. Conversely, the opposite side of the road experiences lower traffic pressure. This project doesn’t appear to be a well thought-out one.”

Deputy Police Commissioner of the Gulshan Traffic Division, Abdul Momen, told the Daily Sun that no one has delegated AI-based traffic management responsibility to the police.

He said, “We haven’t been granted access by anyone. We became aware through the media that DNCC introduced a system. A timer panel was installed in June, but by July, it malfunctioned and remains inactive. Managing traffic solely with traffic lights, without timers, proves challenging for the police, as drivers are unfamiliar with this setup in the capital.”

Offering suggestions for success, Momen remarked, “Implementing the system at a single intersection in the capital isn’t sufficient. If DNCC were to oversee entire corridors, it could yield positive traffic control outcomes. A comprehensive approach is necessary for success, not fragmented steps.”

He said that they requested the DNCC many a times for activating the timers but all have fallen on deaf ears.
On 7 July last, DNCC Mayor Atiqul Islam announced the initiation of the artificial intelligence-based traffic system on a trial basis for the preceding month, reporting approximately 3 lakh signal violations during that period.

A month later, on 7 August, a tender advertisement was issued for the supply and installation of traffic signal controllers. The DNCC purchased supply traffic signal controllers, AI traffic cameras, and e-police cameras as per the advertisement.

A DNCC official, speaking on condition of anonymity, disclosed that the AI-based traffic system was initially introduced as a one-month model project but has continued for eight months. The project was implemented rather hastily without consultations with relevant parties, including traffic police, the official added.

DNCC CEO Mir Khairul Alam acknowledged the need for policy decisions regarding police access to the AI system, stating, “The AI system is currently operating on a pilot basis, providing data on law violators. For enforcement, police require access. Policy decisions on granting access to police need to be made by the home ministry.”
However, Alam could not provide a timeline for granting access to police or for concluding the piloting project.