Cameras on city buses that log changed signs, markings and potholes each week
TL;DR Signchange puts small cameras on city buses that already run the same routes every day. The software compares each week's pass with the last one and flags what changed - a speed sign gone, a faded zebra crossing, a new pothole, a diversion. Road owners pay for the weekly change report and map companies license the feed.
India's road network is owned in bits and pieces. Of roughly 55 lakh km counted in 2015, rural roads made up 61%, PWD roads 20% and urban roads 9%, and national highways were only 2%[1]. Each owner keeps its own records, and PRS notes that road safety duties are spread across bodies with no working coordination[1].
So who has a current list of which signs stand where? Nobody. Survey-grade maps exist, but they're made by fitting lidar, cameras and GPS to dedicated cars[2], and no city can pay for that every week. A missing speed sign can stay missing for months because no one's job is to notice.
What it does, in regulatory terms: road-asset observation and map data only. Signchange collects street-level images under India's 2021 geospatial guidelines, which removed prior approval for most map data[6]. Faces and number plates are blurred on the device before upload, because the DPDP Act applies to them[8]. Check the current geospatial rules for street-level data with a lawyer before launch.
A camera unit sits behind each bus windscreen. It runs a small model that spots signs, lane and crossing markings, potholes and work zones, and it sends only short records plus a cropped image, much like one chipmaker's crowdsourced map sends about 10 KB per km[3]. A bus passes the same spot dozens of times a week, so a change counts only after several passes agree. That's important, since single-pass road detection is noisy[5]. A reviewer checks every flagged change before the customer sees it.
The original version of this idea sold real-time HD maps to driverless cars. India has no driverless fleet to buy them, so I'd drop that. The same pipeline sells today to road owners and to map makers who want fresher data between surveys.
The first payer is a municipal corporation, smart-city company or state PWD that maintains the roads the buses use. They pay an annual fee per route-km for a weekly change report and a ticket list for contractors. The second payer is a map company that wants a change feed to plan its own survey cars. Bus operators get a share per bus for hosting the camera. There's no public price for this kind of service, so working out the price is part of the pilot.
The 2021 geospatial guidelines let Indian companies collect and sell most map data without a licence[6]. PM E-DRIVE is funding 14,028 electric buses for city transport, and each one must carry a monitoring device[9], so new fleets turn up already wired for telematics. On-device vision has also made it cheap to upload only what changed.
There's no public figure for road-asset change data in India. The closest signal is that two listed Indian map companies sell HD and ADAS map products built from survey vehicles[2][7]. One open imagery platform shows the crowdsourced side, with more than 3 billion street-level images from ordinary cameras[4]. Treat both as adjacent and not as your market.
One listed map company builds HD maps with its own multi-sensor survey kit[2]. The other offers lidar mobile mapping and HD/ADAS maps[7]. A driver-assist chipmaker crowdsources map updates, but only from cars with its chips fitted at the factory[3]. The open imagery platform collects images with no service attached[4]. I found nobody in India who turns repeated bus passes into a weekly, verified change list for the agency that owns the road.
| Difficulty | Hard (56/100) |
| Starting capital | ₹1.6–2.9 Cr |
| Time to launch | 9–12 months |
The detection models are known work. The real pain is getting bus operators to let you mount hardware, being accurate enough that engineers trust the tickets, and finding a government buyer who'll pay for data and not just for roadwork.
No mapping licence is needed for most data since 2021, but you'll need written consent to put hardware on buses and a lawyer's check on street imagery.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
Don't send a change to a customer on the model's word alone. A wrong map update on a safety sign is worse than a late one, so a reviewer checks every flag.
Want to work out the AI layer for Signchange? Talk to Yash →
I'd back a founder who has sold to municipal bodies or worked in map production and can sit through a long first sale. If you're hoping to sell to self-driving car makers in India soon then this isn't it. Prove one thing first - on 20 buses in one city, confirmed changes are right at least nine times in ten.
Plan for roughly ₹1.6–2.9 Cr, mostly on bus camera units, the detection software, reviewers and a long first public-sector sale.
Most map data needs no prior approval since the 2021 guidelines. You still need DPDP compliance, bus operator and authority consent and a GeM listing or tender to sell to government.
Yes, road owners and map makers everywhere want fresher data. Check local mapping rules and how strict privacy law is on street imagery first.
The free Signchange validation kit has the line-by-line capital breakdown, the effort scoring, the geospatial and data-protection checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for road owners and bus operators, and 5 assumptions to test. Download the kit →
Thinking about building Signchange or something close to it? The make-or-break work is turning a free pilot with one city into a paid contract that renews, and I help founders with exactly that. If you need a hand, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.