Is India building ADAS cars faster than the ecosystem around them?
Circa mid-2020, COVID was still around, and my wife decided to buy a car for herself because it was unsafe to travel to work in a cab. I still remember when she picked up her Hyundai i20 from a dusty lane in Whitefield, Bangalore.
Coming from a decade-old Mahindra, the i20 felt almost space-age to me: dual front airbags, ABS with EBD, rear parking sensors with a camera and adaptive guidelines, a speed-alert system and seatbelt reminders.
At the time, these felt like significant advances in vehicle safety.
Fast-forward six years to 2026, and the conversation has changed dramatically.
We are now seeing vehicles equipped with SAE Level 2 systems, with features such as Adaptive Cruise Control (ACC), Autonomous Emergency Braking (AEB), Lane Keep Assist (LKA), Blind Spot Detection and Driver Monitoring becoming increasingly mainstream.
And then there are the homegrown heavyweights with sophisticated sensor architectures.
The Tata Sierra EV, for example, has a sensor suite comprising one forward-facing radar, one front-vision camera and four cameras for surround view.
The Mahindra BE 6 (Pack 4) takes a very different approach, with five radars: one at the front, two front-corner and two rear-corner, along with a front-vision camera, four cameras providing a 360-degree view and an in-cabin Driver Monitoring System (DMS).
This enables features such as Summon Mode, Remote Parking and Park Assist.
Pure technical wizardry. All of this has happened in the span of just six years.
The need for these technologies is difficult to ignore. India recorded 1.83 lakh road fatalities in 2025.
But while ADAS technology is advancing rapidly in India, it raises a more fundamental question:
Is the ecosystem around the vehicle advancing at the same speed?
Because an ADAS system does not operate in isolation. Its performance depends not only on the vehicle's sensors, software and computing power, but also on the road, human behaviour and the legal framework governing the system.
And this is where India's ADAS story becomes much more complicated.
Most of the systems currently entering the Indian market are SAE Level 2.
While there are some manufacturers describing their systems as Level 2+, at SBD, we follow the SAE definitions and distinguish systems based on their actual capabilities rather than their marketing terminology. Only those Assisted Driving system with Hands off, eyes on capability are Level 2+
For convenience, here in a classification of ADAS features as explained in detail SBD: 534 ADAS Guide.
But let's be clear these modern Level 2 systems are sophisticated.
They can provide sustained steering and speed control and significantly reduce the driver's workload.
But the driver remains responsible for monitoring the road and supervising the system. Performance of ADAS system through are influenced by the following 3 variables:
The road
The map
The legal framework governing what happens when the system gets it wrong
In India, all three are evolving, but arguably not at the same pace as the vehicle technology.
1. The car may be ready for higher level of autonomy. Is the road?
Consider something as fundamental as Lane Keep Assist or Lane Centring.
These systems need to understand the lane environment. Road markings, lane geometry, curvature, signage and road boundaries all contribute to that understanding.
Now consider the reality of Indian highways.
Lane markings can disappear.
Road layouts can change.
Construction can temporarily alter lane configurations.
Two-wheelers, pedestrians, animals and vehicles can enter the vehicle's predicted path in ways that are difficult to model using conventional highway assumptions
Add high traffic density, aggressive traffic cut-ins and unpredictable interactions between different road users, and the operating environment becomes considerably more complex.
This is also why some drivers may choose to switch ADAS off. False alarms, unnecessary braking or unexpected interventions can quickly erode driver confidence in the system.
For example, the Renault Duster provides a dedicated hotkey that allows the driver to switch off pre-configurable ADAS function with a single press.
This highlights an important point: An ADAS system can be technically capable and still struggle to deliver value if the driver does not trust it in the local operating environment.
And that brings us to an important question:
How do we make ADAS genuinely Indian?
India is already working on this.
Through the SANGARH project - Sangrah: Synchronously ANnotated Ground-truth Acquired & Harvested India Data for ADAS - ARAI has been capturing traffic patterns, object types, road conditions and environmental variations.
The objective is to create a growing, anonymised database that can serve as a foundation for manufacturers and developers building India-specific ADAS algorithms.
But the road itself is only half of the problem.
The vehicle also needs to understand the road digitally through maps.
2. The second problem: the map is becoming part of the driving system
HD maps are often misunderstood as simply more detailed navigation maps.
They are much more than that.
An HD map can provide the vehicle with a digital representation of road geometry, lane topology, speed limits, landmarks and localisation references - effectively giving the vehicle a look-ahead view of the road beyond what its sensors can immediately perceive.
However, without clear physical lane markings, its effectiveness is significantly reduced and can become ineffective in reliably determining the vehicle’s lane position.
India is making significant progress here.
Companies such as MapmyIndia and Genesys are developing lane-level HD datasets for ADAS and automated driving.
In December 2025, Genesys announced an HD-mapping initiative covering more than 100,000 km of national highways, expressways and strategic corridors. Against India's 146,572 km national-highway network, that is roughly two-thirds by length.
Sounds impressive? There is, however, an important catch.
The question is not simply:
“Are India's highways mapped?”
The more important question is:
“Are they mapped, validated and maintained as part of a specific vehicle's Operational Design Domain?”
Consider the US.
The US Interstate System is approximately 78,600 km long. Ford claims BlueCruise covers more than 130,000 miles of pre-qualified divided highway, while GM says Super Cruise covers approximately 750,000 miles across the US and Canada.
These figures are not directly comparable with India's 100,000-km mapping dataset.
The important distinction is that the US figures describe OEM-specific operating domains - roads that have been mapped, validated and approved for a particular hands-free driving system, supported by fleet data and in-service experience.
India's 100,000-km figure, by comparison, primarily describes a mapping dataset that can be licensed across industries.
A map becomes part of an Operational Design Domain only when the vehicle manufacturer validates its system against that map.
And even that is not enough.
The map has to remain fresh.
India's roads can change faster than conventional survey cycles. Work zones appear. Diversions are introduced. Lane markings are removed or repainted. Road geometry changes.
A map that was accurate six months ago may not accurately represent the road today.
That creates an entirely new requirement: map freshness and integrity.
Today, there is no comprehensive national framework defining HD-map accuracy, freshness and provenance for automated-driving applications.
And this brings us to the third and perhaps most complicated piece of the puzzle.
The law.
3. The third problem: the law still sees a human driver
India's Motor Vehicles Act, 1988 places the human driver at the centre of the legal framework.
Section 3(1), for example, requires a person driving a motor vehicle in a public place to hold an appropriate driving licence.
Section 2(9) defines a "driver" in terms of a person, while Section 2(45) defines road markings and related infrastructure as devices intended for the information, guidance or direction of drivers.
That framework made perfect sense when the driver was unquestionably the entity controlling the vehicle.
But what happens when the vehicle itself increasingly makes decisions about steering, braking and acceleration?
Let's consider a hypothetical scenario.
A car with a highly advanced Level 2 software stack is travelling on NH-44 between Bangalore and Hyderabad.
ACC is engaged.
Suddenly, a stray dog runs across the road.
The vehicle detects the dog and activates AEB, while also making an evasive manoeuvre to avoid it.
The manoeuvre successfully saves the dog.
But the sudden intervention causes a tractor carrying a load of sugarcane to brake sharply. The tractor overturns. The sugarcane load falls onto two motorcycles, and a roadside shop is damaged.
Now comes the difficult question:
Who is responsible?
The driver?
The vehicle manufacturer?
The ADAS supplier?
The software developer?
The mapping provider?
Or some combination of them?
Let's make the scenario even more complicated.
Suppose the vehicle received an OTA software update shortly before the incident. The update changed some aspect of the vehicle's driving behaviour, but the driver was not fully aware of the change.
Who is responsible now?
This is where the existing legal framework begins to encounter the realities of software-defined vehicles.
The Motor Vehicles Act introduced Section 2B through the 2019 amendment, allowing the Central Government to exempt certain types of mechanically propelled vehicles from provisions of the Act to promote innovation, research and development.
But an exemption is not a complete regulatory framework.
Type approval is not the same as automated-driving approval
India has already introduced important ADAS requirements.
MoRTH's GSR 834(E), for example, mandates a range of safety features for specified heavy vehicles, including Advanced Emergency Braking Systems under AIS-162, driver drowsiness or attention warning under AIS-184, lane-departure warning under AIS-188, blind-spot information under AIS-186 and moving-off information under AIS-187.
These are important steps.
But these requirements focus on warning and braking performance. That is different from approving a system that continuously steers and controls speed within a defined domain.
The latter requires a much broader framework.
Coming back to our example of the incident that happened on NH-44 highway, it is important to acknowledge that Section 140 of the Motor Vehicle Act, imposes no-fault liability on the owner or insurer, so victims are compensated regardless in the event of an unfortunate accident.
But the harder question is who ultimately bears the loss.
That sits outside the Act currently
There is also an evidence gap.
Section 136A of the Motor Vehicle Act provides for electronic monitoring through mechanisms such as speed cameras and CCTV, primarily to attribute traffic violations to registered owners.
But an increasingly automated vehicle requires a different type of evidence.
Was ACC engaged?
Was Lane Centring active?
Was the driver looking at the road?
Did the system issue a takeover request?
Did the driver respond?
Was a recent OTA update installed?
These could be critical questions in determining what actually happened during an accident.
The US provides an interesting comparison.
NHTSA's Standing General Order 2021-01 requires manufacturers and operators to report certain crashes involving vehicles equipped with Level 2 ADAS and Levels 3–5 automated-driving systems. It specifically mandates reporting if the automation system was active at any point within 30 seconds before the collision occurs. This order helps safety regulators identify technology defects and bridges the evidence lag regarding whether the human or the vehicle software was in control during an accident.
Such a framework provides regulators with an important source of real-world evidence about the performance of automated-driving technologies.
The world is already writing the next chapter
On June 24, 2026, UNECE's WP.29 adopted the first regulatory framework for fully driverless automated driving systems. It requires an audited safety management system, credible testing, a safety case and continuous in-service monitoring and reporting. The benchmark is performance at least at the level of a competent and careful human driver.
That is significant because it recognises that automated driving is not a feature, but required a framework covering the entire operational lifecycle of the system.
India is also moving in this direction.
The country has introduced and proposed measures covering ADAS, V2V communication and intelligent transportation systems. The Department of Telecommunications has moved to exempt parts of the 5.9 GHz band from licensing, while MoRTH has proposed AIS-230-compliant C-V2X units on new L, M and N category vehicles from October 1, 2028.
So the issue is not that India is doing nothing.
The real issue is whether technology deployment, road infrastructure, digital infrastructure and regulation are developing as one connected ecosystem.
The uncomfortable question and the opportunity for India:
Perhaps the biggest mistake would be to treat ADAS as a collection of individual features.
AEB, ACC, LKA, Blind Spot Detection, Driver Monitoring, Traffic Sign Recognition, Highway Assist etc are not just independent features. Rather they are part of an integrated driving system.
That would change the regulatory question of “Does this vehicle have Adaptive Cruise Control?”, to “Under what conditions is this system permitted to control the vehicle, and how has that behaviour been validated for those conditions?”
Further India does not necessarily need to copy the US or Europe as our road condition are different so also vehicle mix, proportion of two-wheelers, interaction between pedestrians, vehicles and stray animals.
India has an opportunity to develop an India-specific ADAS regulatory and validation framework rather than simply importing assumptions from other markets.
India is already building pieces of this ecosystem. The question is whether those pieces will eventually become one coherent framework.
A small personal side note
I guess you are wondering what I drive now. Thirteen years on, I still drive that 2-tonne truck with zero airbags and no ABS. I have, however, opted for an aftermarket rear parking camera with ultra-sonic sensor, for the safety of others.
"India is at an interesting inflection point, with ADAS technology advancing faster than the ecosystem supporting it. The challenge is no longer simply what the vehicle can do, but whether our roads, HD maps, digital infrastructure and regulatory framework are ready to support it. Herein lies the opportunity: to build an ADAS framework designed around India's unique road and traffic environment, rather than simply replicating US or European models." Sasangka Sarmah, SBD Automotive – Senior Consultant |  |
How SBD can help
If you or your team would like to understand how the highlighted technologies, partnerships and market dynamics could impact your strategy, we invite you to get in touch. Email info@sbdautomotive.com to connect with one of our experts and discuss your specific requirements. |
