Walking guidance for blind people, in steps.
Shravan is a phone-first guide for blind walkers in India. It sees the path, says what to do in steps, and checks the step was taken. Stage: design done, prototype next. We are looking for mentors, incubation support and one partner institution.
It sees the crossing.
A camera at chest height reads the kerb, the stripes and the car that has stopped.
Three reasons to back Shravan now.
Why here: the team is in Gandhinagar, and four institutions that work with blind walkers are within an hour. Phase 1 can run close to the incubator, with the people who will use it.
The science and the buyer are both ready.
Harvard showed in August 2026 that a phone app can guide a blind walker outdoors. ADIP's revised limit (September 2024) pays up to Rs 15,000 for an aid, and our kit is planned under it.Proven on this exact problem, with a delivery partner behind it.
First place at Eureka 2026 (IIT Bombay) and selected for SIH 2026. Engineering support from KNTRL, which ships software for NFSU, Interpol and the Government of Gujarat.The buyer already exists.
NGOs, blind schools and ADIP buy aids for blind people today. We sell through them, one institution at a time, so growth does not wait on retail.Every journey is a series of decisions. No tool closes them.
A blind walker in India carries several tools. Each one knows part of the picture. None of them knows whether the walker acted on what it said.
White cane
Feels about a metre ahead. Cannot tell road from footpath.
GPS app
Knows the map. Cannot see the pothole, and speaks in metres.
Vision app
Describes the scene. Does not decide where to step.
A stranger
Helps when asked. But the walker has to ask, and then wait.
A map speaks in metres. A walker counts steps.
No tool checks whether the instruction was carried out.
Turn-by-turn, continuous path guidance, obstacle avoidance.
The three must-haves named by 112 blind and low-vision people in Harvard's survey. None of the aids priced in our market section does all three.One instruction at a time, then a check.
A phone and one earbud. Shravan sees, says the next move in the walker's own steps, and the phone's motion sensors confirm the move before it speaks again. If the walker drifts, it corrects. The cane stays.
Try: play the loop, then turn on "Walker drifts right".
One instruction at a time
-
01 · DetectCamera
Pothole ahead, left side clear
2.2 m away. Shravan turns that into steps.
-
02 · SpeakEarbud
Says what to do
"Pothole, three steps ahead. Space on your left."
-
03 · Walker actsMotion sensors
Steps left
One step, then straight on.
-
04 · Phone confirmsCamera again
Left is clear. Continue.
Checked, not assumed.
| Tool | Sees the obstacle | Says it in steps | Checks it was done |
|---|---|---|---|
| GPS app | No | Metres | No |
| AI describer | Yes | Describes the scene | No |
| Buzzing wearable | Yes | Vibrates | No |
| Shravan (design target) | Yes | Yes | Yes |
- WalkReads the surface ahead, finds free space, gives one instruction in steps, then checks it worked.
- NavigateVoice destination and continuous course correction, instead of "recalculating".
- Recognise"Priya, four steps ahead, slightly right." Only for people the user has enrolled.
- ProtectDetects a fall and asks "Are you okay?" With no answer, it alerts trusted contacts. Never the police on its own.
- TravelBooks a ride by voice, tells the driver, sends family a live link.
- EndureSteps, route, fall detection and SOS work offline. At low battery it sheds intelligence, never safety.
Harvard has shown a phone can guide a blind walker.
Published in Nature Biomedical Engineering in August 2026. Their results, not ours. We build on them for Indian streets.
Try: panels a to d (arrow keys work), then "What it means for Shravan".
Improving outdoor navigation for people with blindness using an AI-driven smartphone application and personalized audio guidance
Harvard John A. Paulson School of Engineering and Applied Sciences
- DOI
- 10.1038/s41551-026-01772-x
- Participants
- 14, from the Carroll Center for the Blind (9 blind, 5 low vision)
- Survey
- 112 blind and low-vision people
What 112 blind and low-vision people said they must have
From the same group's survey. None of the aids we found on sale in India does all three. Shravan is designed around them.
- Turn-by-turn directions Design target
- Continuous path guidance Design target
- Obstacle avoidance Design target
It started as a pitch. We want to build it.
A student team from Gandhinagar, with an engineering company behind it that already ships software for government and institutions.
- 1st place, Eureka 2026A pitching competition hosted by IIT Bombay.
- Selected, Smart India Hackathon 2026IAR internal round. Problem statement 26198, MedTech theme.
- StatusConcept, in design. Not yet built, and not yet tested with blind users. That is what we want to change first.
- Jeet ChauhanPoint of contact
- Anuj BhadouriaTeam Git Ke Ghode
- Arya PatelTeam Git Ke Ghode
Engineering support: KNTRL
A Gujarat technology company that builds software, AI and secure systems. Its record:
- BNIS, a disinformation-detection platform for NFSU and Interpol.
- Inventory, tracking and CRM for SAPTI, Government of Gujarat.
- The AI pipeline that runs a national news channel's desk, from sources to an anchor script in Hindi.
- Rs 2.5 lakh of client work delivered in its first six weeks.
Who owns what: the team owns the product, the research and the work with users. KNTRL builds and secures the software with them.
See KNTRL's work at kntrl.inThree films we will shoot in Phase 1.
We have no footage yet, because nothing is built. Each film plays now as a short motion study of the exact shot, with the shot list under the player. Real clips replace them as we film.
Try: play, scrub, switch film.
Read the films as text
- The walk (40 s, chest-height view at dusk): "Pothole, three steps ahead." "Space on your left." "Left is clear. Continue." "On route. Gate in twelve steps." "Crossing on your right. Wait for my go."
- What you wear (20 s, concept render): chest camera, sees the path at walking height. Open earpiece, speaks without blocking the street. Foot clip, feels each step.
- Phase 1, Gandhinagar (30 s): months 0 to 6 with 20 to 30 walkers and one partner institution. Then Gujarat, 200 to 500 users at 3 to 5 institutions. Then national, with ADIP empanelment.
A large need, reached one institution at a time.
In this market the user is often not the buyer. A blind walker adopts a device from a mobility instructor they already trust, and ADIP, an NGO, a blind school or a state programme pays.
- 340Mpeople worldwide with blindness or moderate-to-severe visual impairmentWHO World Report on Vision, 2019
- 4.8Mblind people in India, a prevalence of 0.36%National Blindness and Visual Impairment Survey 2015 to 19, AIIMS for MoHFW (Vashist et al., PLOS ONE, 2022)
- ~1.5Mindependently mobile, using a smartphone, of working ageOur estimate
- 10Kusers in years one to three, reached through institutionsOur target
WHO's wider figure of 2.2 billion includes uncorrected near vision and is not this market.
The walker.
A mobility instructor the walker already trusts.
ADIP, an NGO, a blind school or a state programme.
| Product | Price | Note |
|---|---|---|
| Torchit Saarthi | Rs 3,500 | Ultrasonic, clamps onto a white cane. Ahmedabad. |
| IIT Delhi SmartCane | Rs 3,000 | Free under ADIP through NGOs. 20,000+ users. |
| Jyoti AI glasses | Rs 12,000 | Describes the scene. |
| Envision Glasses | $2,499 | Describes the scene. |
| OrCam MyEye 3 Pro | $4,490 | Reads and recognises. |
| Meta AI glasses | From $299 | General-purpose AI glasses. |
| Shravan Kit | Rs 7,500 to 12,999 | Planned. Inside the ADIP limit of Rs 15,000. |
The capability exists. The continuity doesn't. Mobilio, Harvard's app, proved a phone can guide a blind person along a path. Shravan is what happens after the path ends: the fall, the ride, the person walking toward you, the phone at 8%.
Free to start. Funded to own.
ADIP (revised 26 September 2024) gives full assistance for aids up to Rs 15,000 to users earning up to Rs 30,000 a month. The kit is planned inside that limit.
Shravan Lite
Free · phone only
Navigation, fall detection, SOS and trusted contacts on the phone the walker already owns.
Shravan Kit
One-time · planned Rs 7,500 to 12,999
Mobility sensor, body camera and bone-conduction earpiece. Inside the ADIP limit, so an eligible user pays nothing.
Software layer
Optional add-on
Core mobility and safety work forever without a subscription. ADIP funds a device, not a software bill.
Institutions
Per seat · annual
Deals with NGOs, blind schools and state programmes, which already buy aids for the people they serve.
One NGO, then five, then a state programme.
We are not selling a device to five million people. Phase 1 is a plan: no user has tried Shravan yet, and no institution has signed.
- Phase 1Months 0 to 6
Gandhinagar
20 to 30 users and one partner institution, from those within an hour: BPA Ahmedabad, Andhajan Mandal, the School for the Blind in Ahmedabad and the Sector 16 blind school in Gandhinagar.
Goal: turn "designed for" into "we asked".
- Phase 2Months 6 to 18
Gujarat
200 to 500 users across 3 to 5 institutions. Outcome metrics published and co-signed by a CRC or DDRC.
Goal: evidence another institution can trust.
- Phase 3Months 18 to 36
National
ADIP empanelment, BIS and ISO certification, and listing as an approved aid.
Goal: change who the buyer is.
Help us build the first working version.
We have the design, the research and the team. In total, Phase 1 needs about six mentor sessions, six months of incubation and one partner institution.
Jeet replies within a day to set up a short call, in person in Gandhinagar or online.
Jeet Chauhan · hello@kntrl.in · WhatsApp +91 75750 37885
- Mentors in assistive technology, hardware and safety.About one session a month through the six months of Phase 1. Unlocks: a prototype that is safe to put in a blind walker's hands.
- Incubation support to build and test a first prototype.Six months, from design to a working kit. Unlocks: trial walks with 20 to 30 users in Gandhinagar.
- One partner institution near Gandhinagar.Starting with listening sessions with blind walkers and their mobility instructors. Unlocks: "designed for" becomes "we asked", and our first outcome data.