Artificial intelligence is helping cities optimize traffic signals, reduce unnecessary stops and improve traffic flow. Los Angeles and Bengaluru show both the potential of adaptive traffic management and the challenges of deploying it across complex urban road networks.
The Technology Is Already in Use
Traffic congestion remains a major challenge in cities such as Los Angeles and Bengaluru. Both cities have introduced advanced systems to improve traffic signal coordination, yet neither has eliminated gridlock.
The problem is not simply a lack of technology. Adaptive traffic signals can respond to changing traffic demand, but their impact depends on how well they are integrated into the wider transport network.
Artificial intelligence can help traffic engineers identify congestion patterns and adjust signal timing. However, it cannot independently solve problems caused by limited road capacity, poorly managed intersections, incidents, construction or excessive demand.
The experience of these two cities illustrates an important distinction: improving traffic flow at individual intersections is achievable, while reducing congestion across an entire metropolitan area requires a much broader strategy.
Los Angeles: A Mature System with Room to Improve
Los Angeles has operated its Automated Traffic Surveillance and Control (ATSAC) system since the 1984 Olympic Games. What began as a system managing 118 traffic signals has developed into a citywide network.
According to the Los Angeles Department of Transportation (LADOT), ATSAC now manages 4,850 traffic signals, supported by more than 26,000 sensors and detectors and approximately 620 traffic-monitoring cameras.
The system allows traffic engineers to monitor road conditions and adjust signal timing as traffic patterns change. It also supports transit signal priority, helping buses and rail services operate more efficiently.
The benefits of computerized signal control in Los Angeles have been documented for decades. An evaluation of an adaptive traffic control deployment at 375 intersections found that travel time fell by 12.7%, average stops decreased by 31% and average delay dropped by 21.4%.
These results demonstrate the potential of coordinated signal control, but they relate to an earlier deployment and should not be interpreted as a measurement of current performance across the entire city.
More recently, LADOT has continued modernizing ATSAC, including through a new control center designed to support more integrated traffic operations.
The remaining challenge is to maintain equipment, improve coordination and make sure that signal optimization serves all road users. Better signal timing can reduce unnecessary delays, but it cannot remove the underlying causes of congestion when traffic demand exceeds available capacity.
Bengaluru: Promising Results, Difficult Implementation
Bengaluru has taken a more recent approach through the Bengaluru Adaptive Traffic Control System (BATCS). The initiative uses traffic data and adaptive signal technology to adjust signal timings according to changing conditions.
The system was launched in May 2024. By October that year, it covered 60 junctions, according to reporting by The Indian Express.
The early results were encouraging. Data shared by Bengaluru’s Joint Commissioner of Police (Traffic), M. N. Anucheth, showed that travel times at Hudson Circle had fallen by up to 33%. Other locations, including K. R. Road and Jayanagar, recorded reductions of up to 20%.
These were reported results at selected junctions, not evidence that travel times had fallen by the same amount across Bengaluru.
The rollout also revealed the practical difficulties of introducing adaptive traffic control at scale. A January 2025 report by Bangalore Mirror described newly installed signals that were still awaiting synchronization. Officials explained that the system needed time to collect traffic data, establish suitable signal timings and coordinate adjacent junctions.
This distinction matters. Installing an intelligent signal is only one step in the process. The system must also be configured, connected to other signals and monitored to ensure that its decisions improve traffic flow in practice.
Bengaluru’s experience shows why implementation quality can be just as important as the underlying technology.
Why Scaling Up Is So Difficult
Several challenges explain why cities cannot simply install AI systems and expect congestion to disappear.
1. Funding and maintenance
Adaptive traffic control requires more than software. Depending on the system, cities may need sensors, communications equipment, upgraded signal controllers, reliable data connections and specialist staff.
The total cost also includes long-term maintenance, cybersecurity, software updates and performance monitoring. Without sustained funding, even a promising pilot can struggle to deliver lasting benefits.
2. Legacy infrastructure and incompatible systems
Many cities operate equipment installed over several decades. Different generations of traffic controllers, communications networks and management platforms may not work together seamlessly.
Integrating these systems can require significant technical work. Open standards and interoperable equipment can help cities avoid becoming dependent on a patchwork of incompatible technologies or individual suppliers.
3. Fragmented responsibilities
Traffic signals, public transport, road maintenance and enforcement may be managed by different agencies. Each can have different budgets, priorities and data systems.
For example, optimizing a signal for private vehicles may conflict with the needs of buses, pedestrians or cyclists. Effective traffic management therefore requires agencies to agree on priorities and coordinate decisions.
Political leadership also matters. Long-term projects need stable funding, clear accountability and the willingness to evaluate new approaches, including pilots that do not deliver the expected results.
4. Data quality, privacy and cybersecurity
Adaptive systems depend on reliable information about traffic conditions. Missing data, malfunctioning sensors or poorly configured equipment can undermine their decisions.
Camera-based systems also raise legitimate privacy questions. Cities need appropriate data-retention rules, access controls and safeguards against misuse. Connected traffic infrastructure must be protected against cyberattacks that could disrupt operations or compromise sensitive information.
These challenges are manageable, but they require ongoing investment and responsible governance.
What Does the Evidence Say About AI Traffic Signals?
Research and real-world deployments show that adaptive traffic management can deliver measurable improvements. However, the scale of those benefits depends on local conditions and the way a system is implemented.
Google’s Project Green Light provides one example. The initiative uses AI and anonymized Google Maps driving trends to recommend changes to existing traffic signal plans. City engineers assess the recommendations before implementing them.
Google reports that early results indicate the potential to reduce stops by up to 30% and greenhouse gas emissions at intersections by up to 10%. These are potential benefits reported by the project, not guaranteed reductions in every city or a promise of equivalent improvements in total journey times.
A separate study published in Nature Communications in February 2025 examined adaptive traffic signals in 100 of China’s most congested cities.
The researchers estimated that replacing conventional fixed-time signals with adaptive systems could reduce peak-hour trip times by an average of 11% and off-peak trip times by 8%. Their analysis also estimated the potential to reduce annual carbon dioxide emissions by approximately 31.73 million tonnes across the cities studied.
Importantly, these figures are modelled estimates rather than the results of a single citywide deployment. Actual outcomes would depend on local traffic patterns, implementation costs and the performance of existing infrastructure.
The evidence nevertheless supports a clear conclusion: adaptive signals can improve traffic flow and reduce stop-and-go driving, particularly when existing signal timing is poorly matched to actual demand.
Why Better Traffic Lights Cannot Solve Every Congestion Problem
Signal optimization is only one part of urban mobility management.
A system may reduce waiting time at an intersection without significantly reducing total journey time if congestion simply moves to the next junction. Similarly, giving more green time to one direction can increase delays for another.
Cities must therefore consider the entire corridor or network rather than measuring success at isolated intersections.
They also need to account for buses, emergency vehicles, pedestrians and cyclists. A system that prioritizes vehicle throughput alone may deliver faster car journeys while making streets less convenient or safe for other users.
In heavily congested cities, demand management, reliable public transport, better walking and cycling infrastructure, parking policy and effective enforcement may be just as important as signal optimization.
AI can help authorities understand these interactions, but the objectives must be defined by transport policy and public priorities.
From Smart Signals to Smarter Cities
Los Angeles and Bengaluru demonstrate two different stages of the same challenge. Los Angeles has a mature traffic-management network that continues to evolve, while Bengaluru’s experience highlights the work needed to install, configure and coordinate adaptive signals across a growing network.
The next step is not simply to install more AI. It is to connect traffic systems, establish clear performance targets and evaluate results consistently.
Cities should measure travel times across entire corridors, intersection delays, public transport reliability, pedestrian waiting times and emissions. They should also publish results so that residents can see whether investments are delivering meaningful improvements.
AI traffic management is no longer just an experimental concept. It is a practical tool that can make existing infrastructure work more efficiently. But its success depends on reliable data, effective implementation and coordination across the transport system.
The real opportunity is not to make every traffic light intelligent. It is to make the entire urban transport network work better together.
