Delhi’s planned Intelligent Traffic Management System is a ₹1,789.52-crore attempt to link adaptive traffic lights, automated enforcement and live travel information across 42 corridors. India’s Cabinet Committee on Economic Affairs approved the project on 30 September 2026. Delhi Police is to implement it in three phases over 24 months, followed by five years of operation and maintenance. The decision approves a system to build; it does not mean the lights or cameras are already operating city-wide.
- The government plans adaptive signal control at 1,092 locations and technology-led enforcement at 1,029 locations across the selected corridors.
- A master system integrator will be chosen competitively by Delhi Police; C-DAC is assigned technical review and project-management support.
- Ambulance priority, quicker incident detection and commuter updates are planned functions, not measured outcomes yet.
- Later phases depend on testing and acceptance of earlier phases. The public approval does not specify a corridor map, vendor, accuracy standard or data-retention policy.
The biggest change is architectural. Delhi already has signals, cameras, officers and traveller information in separate places. The Cabinet’s plan is to make them operate as one network through command centres, a data centre and shared technical interfaces. That could help a busy junction react to traffic rather than keep the same cycle regardless of conditions. It could also make enforcement more pervasive. Whether the project improves journeys without creating inaccurate penalties will depend on procurement, testing, oversight and transparent operating rules.
What did the Cabinet approve?
The Ministry of Home Affairs’ 30 September Cabinet release gives the approved project cost as ₹1,789.52 crore, inclusive of taxes. It names Delhi Police as implementing agency and the Centre for Development of Advanced Computing, or C-DAC, as project-management consultant for technical scrutiny, monitoring, validation and acceptance. It calls for a master system integrator to be selected through competitive bidding.
The release describes 42 identified corridors, not the whole road network at once. Its rollout is planned in three phases over 24 months, with five years of subsequent operation and maintenance. Each later phase is to follow evaluation of the previous one against testing, commissioning, acceptance and performance requirements. The publicly available statement does not name the 42 corridors or say when a procurement notice will be issued.
Hindustan Times, The Times of India and NDTV Profit separately reported the Cabinet decision. Hindustan Times traced the proposal’s growth from an earlier, narrower Central Vista idea. Times of India described a much older effort to create an intelligent traffic system before the 2010 Commonwealth Games. Those histories show why the new approval matters, but neither proves the current plan will be delivered on schedule.
How would an Intelligent Traffic Management System work?
An Intelligent Traffic Management System, or ITMS, uses current road conditions to adjust control and information rather than relying only on fixed schedules and isolated observations. The Cabinet announcement proposes adaptive traffic-control equipment at 1,092 locations. Such a system can lengthen or shorten signal phases according to measured traffic and coordinate nearby junctions along a corridor. It does not create new lane space, and it cannot guarantee a congestion-free commute.
At 1,029 other specified enforcement locations, the plan includes technology-based detection of violations. The official list includes red-light and stop-line violations, speeding, wrong-way driving, missing helmets or seatbelts, illegal parking and mobile-phone use while driving. AI-based analytics are also proposed to help identify stolen, hot-listed or blacklisted vehicles, repeat violators and vehicles connected to hit-and-run cases. These are intended capabilities, not evidence that each category has been tested or that every machine-generated flag will justify a penalty.
The Cabinet also describes a city-wide command layer. Traffic Command and Control Centres would bring together regulation, enforcement and traveller information, backed by a secure data centre and disaster-recovery infrastructure. Open standards and APIs are promised so the platform can integrate additional systems later. A standard interface may prevent vendor lock-in, but the release does not publish the actual interface specifications, ownership terms or access controls.
Why link signals, enforcement and information?
Each function can operate alone, but an isolated system sees only part of a traffic problem. A fixed-timer signal cannot respond to a sudden diversion. A camera may record a breakdown without influencing the next junction. A travel message can become stale if it is not fed by live road information. The ITMS proposal tries to connect these actions so information from one part of the network can change a decision elsewhere.
The ministry says Delhi has around 71 lakh active registered vehicles, while road growth has lagged vehicle growth. That figure is a government description of the problem, not a promise that better software can substitute for every road or transit investment. The project might help make better use of existing capacity, yet queueing can simply shift from one junction to another if neighbouring intersections are not coordinated and transit alternatives are overlooked.
Hindustan Times notes that some Indian cities already operate adaptive signals. The new Delhi project is therefore not a claim to have invented AI traffic lights. Its proposed scale and integration across 42 corridors, 1,092 control locations and 1,029 enforcement locations are the news. The operational challenge is connecting equipment and agencies reliably, especially when existing devices may have different owners, data formats and maintenance contracts.
Delhi’s physical network is also changing through road projects such as the approved Dwarka tunnel plan. A traffic-control platform is a different intervention: it manages the use of existing streets and the information drivers receive. It will still need updated road maps and signal plans as new links open, junctions change or construction reroutes traffic.
What will commuters actually notice?
If deployed as described, a driver might see a signal whose timing changes with conditions on that corridor rather than repeating a fixed cycle. Drivers may also get live congestion, diversion and parking information through variable-message signs and digital channels. The government says information may be shared with navigation providers, but it does not specify the recipients, data format, delay or whether a private mapping service has agreed to use it. Lapaas Voice has covered separate India updates to Google Maps; those features should not be conflated with an ITMS integration that has yet to be announced.
For an ambulance, the proposal is to support priority movement through a green corridor. That is one of the most important potential benefits, but it requires more than a camera spotting a vehicle. The system must receive a trustworthy emergency request, know its route, change signals safely, alert operators when needed and release priority once the vehicle passes. The Cabinet statement does not detail this workflow or provide a target reduction in ambulance travel time. Claiming a specific lifesaving benefit today would go beyond the evidence.
For a motorist, automated enforcement could become more visible. The ministry lists a wide range of detectible violations, but the difference between a machine flag and a lawful fine is crucial. A camera can misread a plate or misclassify a manoeuvre; weather, occlusion and night conditions can affect evidence. The government’s release does not specify accuracy thresholds, appeal procedures, review by a human officer or error reports. Those details matter to fairness and public trust and should appear in the tender and operating policy.
How large is the rollout?
The Cabinet’s counts describe locations, not individual cameras or a guaranteed one-to-one mapping. One site could involve multiple devices; a single corridor could contain multiple junctions and enforcement points. Nor does the statement give the length of the 42 corridors. A reader should not multiply these figures to invent a city-wide camera total or assume every Delhi junction is included.
| Measure | Cabinet figure | Status |
|---|---|---|
| Estimated project cost | ₹1,789.52 crore | Approved estimate, including tax |
| Selected traffic corridors | 42 | Planned coverage |
| Adaptive signal locations | 1,092 | Planned control sites |
| Enforcement locations | 1,029 | Planned detection sites |
| Implementation | Three phases over 24 months | Subject to testing and acceptance |
| Operations and maintenance | Five years | After rollout |
Who may win the technology work?
Delhi Police is expected to procure a master system integrator by competitive bid. That bidder would have to bring together hardware, software, communications and operations according to the final tender. The ministry assigns C-DAC a separate role: technical scrutiny and validation as project-management consultant. The approval does not name a winning vendor, subcontractor, cloud operator or camera maker. Any company-specific revenue forecast at this stage is speculative.
The project could attract businesses that build traffic controllers, edge-video systems, data-centre infrastructure, security tools, signs, network connections and analytics. The requirements may also create work in interoperability testing and maintenance. Yet an attractive budget is not the same as revenue. Contract value and profit depend on tender scope, obligations over the five-year maintenance period, penalties for downtime and the cost of integrating legacy traffic equipment.
The promised open standards and APIs will be a particularly important procurement test. If published interfaces let different suppliers’ components be changed without rewriting the entire platform, the city could preserve competition and avoid being tied to one vendor for every future upgrade. If the contract simply says “open” without enforceable documentation, testing rights and data portability, the promise may have little practical effect. The Cabinet release gives the intent; bidders and civil-society observers will need the detailed specification to judge it.
What could go wrong with AI enforcement?
Automation can provide more consistent observation than a human standing at one junction, but it also scales mistakes. A false flag can become a mass problem when many cameras share the same model or rule. A reliable intelligent traffic management system needs performance tests for different light, rain and road-marking conditions, plus a way to review and correct bad evidence. The public statement does not set those thresholds, so the eventual tender should be read closely.
Privacy is another unresolved implementation issue. Detecting a violation or locating a stolen vehicle can require image capture, number-plate processing and cross-checks with other databases. The Cabinet release mentions a secure data centre and disaster recovery, but gives no retention period, access log policy, independent audit process or limits on secondary uses. These are questions for the project documents, not grounds to assume a particular surveillance design has already been approved.
There is a useful caution in Lapaas Voice’s earlier reporting on public backlash against AI traffic cameras in Mississippi. That foreign episode does not determine Indian law or Delhi’s procurement, but it shows why a road-safety programme needs clear enforcement rules and public trust. Delhi can establish those rules before deployment instead of relying on a technology label to answer them.
Operational resilience matters just as much. If a central connection fails, junctions must still operate safely. If clocks drift or a database is unavailable, enforcement records need an auditable fallback. If an emergency vehicle receives priority, the system must prevent unsafe conflicting movements. These are engineering requirements to test, not claims that the approved platform already meets them.
What should Delhi publish next?
The first milestone is a corridor list and phased schedule. Residents should be able to see which junctions are in Phase 1, how existing signals will transition and when each segment is tested. A map would prevent the assumption that every road is covered. It would also let the city collect useful baseline measures—queue length, travel time, crash rates and emergency-vehicle response—before changing timings.
The second milestone is the competitive tender. It should clarify who owns data and software, who can inspect model decisions, how errors are appealed and which functions require human review. It should also separate planned features from guaranteed deliverables. An “AI-enabled” label is not a test result. Measurable acceptance criteria would make the three-phase gate meaningful.
The third is public reporting once segments go live. Time saved at one intersection can be lost at the next; a corridor-wide measure is more informative than a demonstration at a single junction. The city should report outcomes for travel time, safety, emergency priority and false-positive enforcement, while explaining changes in roadworks and demand. Without such evidence, the ₹1,789.52-crore approval remains a spending plan rather than demonstrated urban improvement.
The decision in one paragraph
Delhi’s Intelligent Traffic Management System is an approved technology project, not a completed deployment. On 30 September 2026, the Cabinet Committee on Economic Affairs approved an estimated ₹1,789.52 crore for Delhi Police to integrate adaptive signals, automated enforcement, incident response and traveller information across 42 corridors. The rollout is planned in three phases over 24 months, followed by five years of maintenance. The government expects safer, smoother travel, but no measured benefit or successful vendor has yet been established in the public approval.
Frequently asked questions
Are Delhi’s AI traffic signals already live across 42 corridors?
No. The September 30 decision approved the project. Its planned three-phase implementation spans 24 months, and later phases depend on earlier testing and acceptance.
How many cameras will the system install?
The government lists 1,029 enforcement locations and 1,092 adaptive-control locations, but its short release does not state an exact total of individual cameras or devices.
Will AI automatically issue every traffic fine?
The official release describes automated detection of several violations. It does not explain the legal review, evidentiary checks, appeal process or error thresholds for issuing a fine.
Who will build and check the system?
Delhi Police will competitively select a master system integrator. C-DAC is assigned technical scrutiny, monitoring, validation and acceptance support. No supplier winner is named in the approval.
Sources: The primary record is the Ministry of Home Affairs Cabinet announcement of 30 September 2026. Its scope and date were checked against separate original reports from Hindustan Times, The Times of India and NDTV Profit. Expectations are identified as government plans rather than independently measured results.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
