New Delhi: In the heart of South Extension’s bridal market, where mannequins in sequin-studded lehengas glitter behind glass, lies a control room. Here, at least 10 young engineers sit, their eyes oscillating between computer screens and nine TV panels flashing live images of traffic junctions from Delhi, Ahmedabad, and other cities.
They are all looking for potholes, that ubiquitous catch-all metaphor for everything wrong with India. After trying to solve the stubborn problem through various means for over seven decades, engineers are now turning to Artificial Intelligence.
This South Ex room is the command centre of Nayan Technologies, an AI road-safety startup. From here, they track every dent in asphalt, every crack in tarmac, through cameras mounted on government buses and police vehicles.
Potholes in India are not just an everyday nuisance but a deadly risk. In 2023 alone, 2,161 lives were lost because of potholes, with UP and MP reporting the highest toll, according to the Ministry of Road Transport and Highways. Just this Monday, a man in Palghar was crushed to death by a tanker after his bike hit a pothole; in MP’s Khargone there were protests when a 26-year-old man died in a similar crash, after which repair work reportedly started. Potholes also turning into a political flashpoint, with the BJP and Congress trading barbs over the potholes in Bengaluru. The BJP even took out a protest on 24 September over the “City of Potholes”, while Deputy CM DK Shivakumar argued that there were “potholes in front of the Prime Minister’s house” as well.
From 'Silicon Valley of India' to 'City of Potholes.' The @INCKarnataka government is systematically destroying Bengaluru's global reputation. Their promise of smooth roads has turned into a daily nightmare for commuters and a symbol of their broken promises. pic.twitter.com/hyTqAPVuLM
— BJP Karnataka (@BJP4Karnataka) September 24, 2025
For the engineers in South Ex, politics is background noise. Their job is to find potholes as quickly as possible.
“A bus can act as a moving CCTV,” said Umang Gupta, head of sales at Nayan. “It catches violations naturally. At a fraction of the cost of an ITMS [Integrated Traffic Management System], which depends on static cameras at junctions. That’s why we are building dashcam tech.”
Across India, a new generation of AI-powered startups is reimagining road safety. RoadMetrics, based in Bengaluru and London, turns smartphone cameras into pothole detectors. Pune-based RoadBounce uses sensors to measure road roughness and sends alerts to civic bodies. Nayan Technologies has rolled out dashcams to flag real-time road defects. And perhaps the most recognisable startup to enter this space is leading EV two-wheeler manufacturer Ather Energy. It’s now collecting pothole data through its electric scooters.
Potholes are the final frontier for a technology that the world is still trying to grapple with.
Until recently, roads were mapped through manual surveys — engineers driving around, clipboard in hand. Then came drones, scanning highways for fractures. In the last five years, however, AI models have been working to predict which stretches may cave in after the next monsoon. And smartphone applications geo-tag broken patches the moment a vehicle runs over them.
These systems are taking over the job of city PWD officers. The information is shared in real time with the city traffic department for action.
Potholes are only the beginning. Live data is our biggest strength. We’re working to harness it by building more sophisticated algorithms
-Swapnil Jain, co-founder of Ather
“AI can detect road hazards in a week, which would take three months [with traditional manual audits],” said Piyush Tewari, founder of SaveLife Foundation, a non-profit working on road safety.
Tewari, however, added a caveat. AI can accelerate information exchange, but faster action requires transparent systems, independent safety budgets, and accountability for contractors, and agencies. “Without that, technology alone cannot save lives.”
India’s road sector is at an inflection point. AI is not just a tool but a “lever for national development”, according to a 2025 KPMG report titled ‘AI-powered Road Infrastructure Transformation – Roads 2047’. The shift towards new technologies to solve the age-old problem of traffic and potholes is already visible.
Delhi is experimenting with AI to maintain its limited stretch of “European standard” roads. The transport ministry has cleared an AI-based safety project in Uttar Pradesh. The National Highways Authority of India has set up a dedicated AI cell to push innovation. And a slew of new-age startups are piloting AI pothole-mapping to traffic flow optimisation.
But in the absence of clear data rules, experts warn that AI algorithms and output must be “explainable and auditable”, and that workforce readiness is equally critical.
Also Read: Road building is a money-making racket in India. And we have a very short memory
From potholes to AI
On India’s rain-battered roads, potholes are an old nemesis. Every monsoon brings them back. Just last week in Patna, a Scorpio SUV sank into a waterlogged crater near the railway station. The owner said authorities ignored the hazard for 20 days, even as she called it a conspiracy to defame the Bihar government. Both the images and the interview went viral.
Not much before this, another pothole-related video went viral—but this one was about a solution. In the first week of September, Ather Energy CEO Tarun Mehta posted a video on Instagram announcing a new ‘pothole alert’ feature on the company’s electric scooters. It got over 2.4 million views.
“I kid you not, there are people in Ather who have been pitching this feature for the last eight freaking years. It’s only now, with lakhs of Athers, enough data, and enough compute power that we are finally able to bring this feature,” he said, explaining the ‘Bluetooth architecture’ behind it and showing a map of Bengaluru dotted red with potholes. Pune, Delhi, and Mumbai too had been mapped, he added.
Over 3,500 comments flooded in. Users tagged municipal bodies, some pleading for collaboration, others mocking civic indifference. “BMC, perhaps use this to fix roads,” read one. “Wouldn’t it be easier to fix the roads?” asked another.
Now technology promises a way out. For Ather, that means turning scooters into moving labs for pothole detection and warning.
Ather, a 12-year-old company from Bengaluru, began as a project at IIT-Madras, where engineering design student Mehta and co-founder Swapnil Jain started out by trying to design better batteries. It later evolved into what they call “India’s first smart electric scooter.”
The pothole feature idea had been brewing since Ather’s first scooter hit the roads. Every vehicle came fitted with an IMU, a sensor module that tracks motion, tilt, and shocks. With half a million Athers now covering over one crore kilometres daily, the company sits on a live, city-scale dataset of road conditions.
The team endured many back-breaking rides to isolate what a pothole looks like in sensor readings. They tested multiple machine learning models and optimised them to run both on the scooter’s edge hardware and in the cloud, said Swapnil Jain, Ather’s co-founder and CTO, and Mehta’s classmate at IIT-Madras.
The system now flags potholes and broken patches with accuracy, categorises them, and issues rider-first cautions. In Bengaluru, Chennai and Hyderabad, where it’s been rolled out so far, alerts are delivered through Ather’s Halo smart helmet. Behind the scenes, 4TB of data is processed every day, feeding a live stream of geospatial and sensor inputs that Ather is working to integrate into traffic management, road audits, and EV infrastructure planning.
Maps should ideally tell us when to slow down. Roads should warn us
-Tabrez Alam, founder of early pothole detector Intents Mobi
“Potholes are only the beginning,” Jain said. “Live data is our biggest strength. We’re working to harness it by building more sophisticated algorithms.”
While Ather’s scooters turn riders into real-time sensors, other startups are using smartphones and AI to map the country’s roads more systematically.
Bengaluru-based RoadMetrics AI charts road conditions with artificial intelligence. Its software tracks defects like potholes, cracks, and ravelling, along with assets such as signals, streetlights, and road markings.
Their goal is to give cities and infrastructure firms the data they need to plan maintenance and budget better. Unlike costly LiDAR or 360° rigs, RoadMetrics uses something more scalable: a smartphone. Inspectors mount phones on their windshields, drive around, and the app captures video and metadata. AI models then process the footage to detect potholes, faded markings and missing signs. The results appear on a GIS dashboard.
“It looks like Google Maps, but with our own road condition layer,” said Dipen Babariya, co-founder of RoadMetrics AI. Each road is graded from green to red—good to poor. The mappers can even click on a detected pothole image and navigate directly to the spot via Google Maps.
The startup works strictly with government clients. It has mapped Jamshedpur with Tata Group, partnered with Mumbai’s municipal corporation, and is digitising Bengaluru with the city’s traffic police. Abroad, it works with 40 councils in the UK and several in Australia.
Back in 2018, Babariya and his classmates, then engineering students at Surat’s Sarvajanik College of Engineering and Technology, built a prototype app that could alert users about road conditions through their smartphones. The idea caught the attention of the municipal corporation, giving them their first foothold in government partnerships.
India, though, still lags in preventive maintenance. A crack today becomes a pothole tomorrow. Fixing it early costs a fraction of repair, he added.
In contrast, the UK follows visual Pavement Condition Index standards, patching defects at the earliest stage. India mostly relies on roughness-based surveys or slow vehicles that scan highways once a year.
The scale is daunting — 6.1 million km of roads, coupled with chaotic traffic, dust, and garbage. AI still struggles in each condition.
“It’s like training a child,” Babariya said. “We have to keep feeding it better data.”
Sometimes, what looks like a pothole is just a pile of trash. So RoadMetrics has trained its models in garbage detection too.
Plugging funding gaps
Behind the shiny dashboards and algorithms is the perennial challenge of funding and scale.
Going by global trends, Babariya has a clear idea of where RoadMetrics is headed: a marketplace where automakers, government agencies, and AI platforms share road-condition data in real time.
For now, the company runs lean, with 15 people split between Bengaluru and London. Investors are scarce in B2G, but RoadMetrics had early backing from VCs and angel investors. The plan is to grow organically from revenue, aim for an IPO, and avoid burning through venture capital.
It’s like training a child. We have to keep feeding it better data
-Dipen Babariya, co-founder of RoadMetrics AI
Ranjeet Deshmukh, founder of Pune-based RoadBounce, took another route. A software engineer for 25 years, he was looking for a new idea in 2019 when a Maharashtra road official aired out a long-standing complaint: “We have the second-largest road network in the world but most of it is old, battered, and invisible to us. We only know when citizens complain.”
The challenge for Deshmukh was not to fix the potholes but to find them.
Using the same sensors that track steps, his team captured vibrations of moving vehicles. They mapped these jolts and measured cracks, bumps, and potholes.
The breakthrough was an algorithm that stripped away the noise of different cars, tyres, and phones.
“That was our USP,” he said.
In Maharashtra, they surveyed 25,000 km of highways in just six weeks. It could have taken years, and crores of rupees, with conventional technology.
As AI matured, they layered in cameras, dashcams, and car sensors. Their maps spread to Puerto Rico, Singapore, Vietnam, and parts of South Asia. Scaling India, however, proved more difficult.
“Singapore is tiny and in order. Indian roads are different terrain,” Deshmukh said.
Consumers remain another challenge. RoadBounce provided data on potholes that Google Maps could not, but giving it away free was unsustainable. The hope now is that giants like Google, Apple, or automakers will plug it in as the missing link.
Bootstrapped since it was founded in 2017, RoadBounce turned down Shark Tank funding, according to Deshmukh. The value isn’t in just the kilometres mapped, he said, but in lives saved.
Unlike RoadBounce, Nayan Technology is scaling its model on VC funds. It was founded by Jayant Ratti, a PhD in Robotics and AI from Georgia Institute of Technology. He spotted an opportunity in common road hazards while he was working as an Uber driver in the US to study road conditions.
In 2019, it relied on crowdsourced data from drivers, before pivoting to installing hardware in public transport and government vehicles. Its dual-channel dashcams capture both the road and the driver’s actions.
This approach attracted significant investment. In October 2024, the company raised $2 million in a pre-series round led by BEENEXT, with participation from We Founder Circle, Venture Catalysts, LetsVenture, and FAAD Capital. The money is being used to expand its AI and IoT solutions in India and abroad. Nayan has secured pilots in over 15 cities, including Ahmedabad, Jalandhar, and Lucknow. The name comes from its goal: to equip vehicles with “eyes”.
Every kilometre counts
Some pothole solutions have emerged from personal urgency. People who experienced the hazards of Indian roads first-hand have built technology to prevent accidents.
In 2018, on Gurugram’s Sohna Road, Tabrez Alam found himself gripping the wheel tighter than usual. His wife, pregnant and struggling with high sugar levels, needed frequent visits to Fortis Hospital in Huda City Centre. Each trip came with sudden jolts and the constant fear of an accident.
“That’s when it hit me,” he said. “Maps should ideally tell us when to slow down. Roads should warn us.”
By 2020, Alam and three others had begun building Intents Mobi, a phone-based system that could detect potholes every time a car jolted.
“When a car experiences a jerk, your phone does too,” Alam said.
Within 18 months, their algorithm could identify road hazards. What began as a personal worry had evolved into a company mapping India’s overlooked crisis. Every day, the app scanned 15 lakh kilometres of roads. By 2023, it had five million users and was processing two billion data points daily.
“On average, the system detected 15,000 potholes, while 12,000 were being fixed by real-time users who updated us,” Alam said. Clusters of potholes formed heat maps, hazardous patches revealed patterns, and Uber and Ola drivers signed up as logistics firms tested the system.
But then came an unnavigable bump in the road. In late 2023, as the startup was raising $5 million, Google Maps slashed prices. Ola Maps launched at the same time.
“Customers ran away,” said Alam, who has shifted to Goa and visits Gurugram once a month for work. “When maps are free, how do you survive?” Investors pulled back. By January 2025, the servers dedicated to pothole detection were shut.
For Alam, an intellectual property consultant by profession, the irony is bitter. He said startups in Germany and the UK secured millions in government grants for road maintenance innovations. In India, there was hesitation from governments.
“We keep copying the West,” he said. “We need to build for India. AI has to help India.”
Millions of kilometres of roads in the country remain unmonitored and dangerous. In 2023, the last year for which official data is available, crashes killed more than 1.7 lakh people in India and left nearly 5 lakh severely injured.
“In almost every crash, there is a human factor, a vehicular factor, and an infrastructure factor at play,” said SaveLife Foundation founder Piyush Tewari.
Among the infrastructure issues, something as ordinary as a pothole can be deadly, but AI can make a difference, he added.
Also Read: Belagavi to Boeing. Karnataka aerospace hub is a shining postcard for Make in India
‘The pothole killed her’
Manual audits crawl along at the pace of Bengaluru’s traffic, but technology can sweep through highways in minutes, flagging cracks, undulations, or gaps in medians before they turn fatal.
But faster detection is the only part of the battle. The bigger obstacle is budgets and how data is acted upon.
Safety gets sidelined because it receives only a sliver of the road budget. In 2025-26, the Ministry of Road Transport and Highways earmarked Rs 595 crore for safety, against more than Rs 2.7 lakh crore for highways development. Even this small allocation is often underused — the ministry, for instance, spent Rs 279 crore of the Rs 330 crore set aside in 2023-24.
There is some progress. The NHAI has begun using drones and AI, and a 2019 amendment to the Motor Vehicle Act cleared the way for automated traffic enforcement and made contractors liable for accidents caused by poor road design or maintenance. However, scale is a challenge. Most roads fall under state agencies — PWDs, municipal bodies, and expressway authorities that have yet to adopt such tools widely.
“Until they do, AI’s full potential will remain untapped,” said Tewari.
Most of our public agencies do not have the technical in-house capabilities that they need. Why not build up in-house capabilities? Train the people you already have. Hire technically savvy people
– Girish Agarwal, professor at IIT-Delhi’s Transport Research and Injury Prevention Centre
For Girish Agarwal, professor at IIT-Delhi’s Transport Research and Injury Prevention Centre (TRIP), who has worked at the intersection of road infrastructure and urban planning, the deeper problem lies within the institutions themselves.
“Most of our public agencies do not have the technical in-house capabilities that they need,” he said.
Engineers in PWDs and municipal bodies spend most of their time on contract paperwork instead of technical assessments. Agarwal suggested that instead of buying costly tools and outsourcing data handling to consultants, agencies should focus on people.
“Why not build up in-house capabilities? Train the people you already have. Hire technically savvy people.”
Meanwhile, there are real lives at stake in the wicked problem of India’s potholes.
Every day for years, 44-year-old nurse Madhavi rode the 12-km stretch from her home in Baikampady to AJ Hospital in Mangaluru. On the morning of 9 September, a rain-filled pothole on NH-66 near Kuloor jolted her scooter. She lost balance, and a lorry carrying fish barrelled into her and crushed her.
What followed was familiar theatre. Police filed an FIR against NHAI officials for negligence. Congress and BJP exchanged barbs about the other’s civic apathy. Residents marched with hand-painted signs declaring “potholes are public murder”. Officials rushed into emergency meetings to explain why repairs were delayed despite a Rs 28 crore tender.
For Madhavi’s husband, Satyashankara CH, the protests and politics have little meaning. But the pothole haunts him.
“If only there was no pothole, she would have been alive,” he said over the phone. “NHAI is responsible. The road killed her.”
(Edited by Asavari Singh)