Impact — 50%
Three-station deployment; 75 enrolled passengers; user support; independent evaluation
Evidence at month 12: Independent route completion, time, assistance, confidence, gender and technical reliability results
ADB Technology Innovation Challenge
BraillePath
Find your way through a metro station. On your own.
- Lead applicant
- E-Sia Learning, Ghent
- Operator partner
- CMRL, Chennai — role confirmed
- DPO co-design partner
- Nethrodaya, Chennai — confirmed
- Project period
- 12 months, expected start January 2027
- Grant requested
- USD 450,000
- Co-financing
- USD 50,000 (10%)
Accessible stations still leave the hardest part unsaid.
Tactile paving gets you into the building. Nothing tells you where to go next.
Overview
Guidance at the moments that decide a journey.
Entrances. Fare gates. Forks. Stairs. Platforms. Interchanges. Exits.
Short instructions through audio, screen reader, refreshable braille and haptics. Everything runs from a downloaded station pack, offline.
A 12-month, three-station pilot in Chennai with 75 blind and low-vision passengers.
Design principle
BraillePath never sends a passenger toward a platform edge on a drifting sensor estimate, and never asks anyone to count steps.
Read the full propositionHide detail
BraillePath will test an offline-first station navigation service that gives blind and low-vision passengers concise, private and self-directed instructions at entrances, fare gates, forks, level changes, platforms, interchanges and exits. It works with the passenger's existing cane or guide dog and with the station's tactile paving and announcements — it replaces none of those safety layers.
The service combines an accessible mobile client, a station route graph, BLE anchors at decision points, phone-sensor motion estimates between anchors, and optional verification markers. Core guidance runs from a downloaded station pack and does not depend on continuous internet access.
The ask is a 12-month, three-station pilot with CMRL and blind users in Chennai, funded through a USD 450,000 ADB grant and USD 50,000 documented co-financing from E-Sia Learning — producing measured passenger outcomes, technical acceptance evidence and a procurement-ready replication package.
Delivery is partner-led: CMRL provides the operating environment, Nethrodaya leads blind-user co-design and local delivery, Prodata Group supports transit-system integration, and CPPR provides independent evaluation. All roles are confirmed for this submission.
Assessment criteria
What the grant buys, and what it proves.
Innovation — 30%
Hybrid native/web product; offline station packs; calibrated BLE/PDR; accessible outputs
Evidence at month 12: A working integrated system that fails safe and is tested in crowded metro conditions
Scalability — 20%
Open, portable data formats; operating model; procurement and replication package
Evidence at month 12: CMRL adoption roadmap plus a costed blueprint for other Indian and ADB-supported operators
A confirmed team, not a wish list.
CMRL, Nethrodaya, Prodata Group and CPPR have agreed their roles for this submission.
Partners
Five organisations. One chain of evidence.
E-Sia does not approve station safety, evaluate its own impact, or stand in for blind-user expertise.
E-Sia Learning
Lead applicant: product, accessibility, instructional design, project management and reporting.
Confirmed — owns BraillePath background IP and provides USD 50,000 co-financing
Chennai Metro Rail Limited
Operator partner: station access, operating requirements, staff interface, safety review and adoption roadmap.
Role confirmed — exact sites, installation plan and public-trial release pass CMRL's site and safety gates
Nethrodaya, Chennai
Paid DPO co-design and local delivery: recruitment, advisory panel, route walk-throughs, language review and coordination.
Role confirmed — inception workplan and participant safeguards completed before recruitment
Prodata Group, Belgium
Transit-systems advisory and integration: positioning architecture, site survey, installation, calibration and staff handover.
Role confirmed — station design and installation method pass CMRL review
CPPR, Kochi
Independent evaluation and policy: baseline, midline and endline, gender-disaggregated analysis and policy translation.
Role confirmed — independence arrangements and protocol approved before deployment
Walkthrough
One journey, screen by screen.
What the passenger hears, feels and reads in braille — including what happens when the system is not sure.
Before travel
01 · Prepare and choose
The passenger downloads the approved, signed station pack and selects audio, screen-reader/braille and haptic preferences. Core guidance works without a connection from this point on.
1 / 10
Station pack
Ready offline
Thirumangalam → Guindy · v1.4 signed
Confidence state
Confirmed
Positive anchor matches the route node
Braille
⠕⠋⠋⠇⠊⠝⠑
Haptic pattern
1 short pulse · pack ready
Offline firstTamil · English · Hindi
Contact station staff
It would rather say nothing than say something wrong.
If confidence and route logic do not agree, the passenger is asked to stop, re-confirm or contact staff.
Technology
Five layers, one fail-safe rule.
Layer 5
Accessible outputs
Short phrases in Tamil, English and Hindi through audio and screen reader, the same text on refreshable braille displays, and a small learnable haptic vocabulary. Critical information is never conveyed by sound alone.
WCAG 2.1 AA targetBraille loan poolTalkBack compatible
Layer 4
Offline-capable native client
An Android-first client (Flutter or equivalent) handles background BLE scanning, inertial sensors, haptics and secure event storage. The web prototype remains the demonstration and pre-trip familiarisation channel.
Signed station packsNo live connection required
Layer 3
Route-constrained confidence engine
The client combines the last verified node, expected next nodes, signal stability, dwell time and inertial movement. Hysteresis prevents instructions from oscillating in crowds or around metal structures.
Four confidence statesNo silent advance
Layer 2
Positive anchors and verification markers
BLE is a zone and identity signal, not centimetre ranging. Each anchor maps to a route-graph node and is calibrated on site. QR markers give exact user-initiated confirmation; NFC is tested at selected points.
20–50 anchors per station6–15 markers per station
Layer 1
Physical safety baseline
Cane or guide dog, tactile paving, station announcements and staff. BraillePath complements these layers and is not a collision-avoidance or obstacle-detection system.
Complementary by designOwned by CMRL
See the four confidence statesHide detail
| State | System rule | Passenger message |
|---|---|---|
| Confirmed | Positive anchor matches the current route node and the stability threshold is met | Concise next instruction in the selected channels |
| Probable | Sensor estimates and the expected route agree, but the next anchor is not yet confirmed | Continue along the tactile path; no safety-critical advance |
| Ambiguous | Signals conflict, are stale or suggest an impossible transition | Stop in a safe position and re-confirm at an approved marker |
| Support | The passenger requests help or cannot re-confirm | Call the approved station contact; share the last verified waypoint only with consent |
Delivery plan
Twelve months. Five gates.
1 · Mobilise and baseline
M1 – M2
Written roles; evaluator protocol; user advisory panel; baseline routes; ethics and privacy; station survey.
Exit gate: CMRL authorisation, signed route scope and baseline complete
2 · Co-design and harden
M3 – M4
Native and offline build; Tamil, English and Hindi content; route graph; marker discovery tests; staff protocol.
Exit gate: Accessibility and design review passed
3 · Closed station test
M5 – M6
Install and calibrate station 1; 15-user test; technical, safety and staff drills.
Exit gate: ≥95% correct triggers, false advances ≤2%, safety panel approval
4 · Full operational pilot
M7 – M10
Deploy remaining stations; 75-person cohort; peak and off-peak observations; monthly DPO review.
Exit gate: Data completeness and recruitment thresholds maintained
5 · Evaluate and transfer
M11 – M12
Endline; independent report; CMRL roadmap; maintenance and procurement package; ADB knowledge output.
Exit gate: Evidence, asset register and replication blueprint accepted
Results framework
What we measure.
≥ 85%
Journeys completed independently at endline
≥ 25%
Shorter median navigation time on matched routes
≥ 45%
Women among active participants — a recruitment gate
Show all seven indicatorsHide detail
A repeated-measures design compares the same participants on matched routes at baseline and endline, disaggregated by gender, level of vision, braille use, device class and station.
| Indicator | Definition and method | Pilot target |
|---|---|---|
| Independent route completion | Correct entrance–platform–interchange–exit task with the usual mobility aid, without route directions or physical escort | ≥ 85% of assessed journeys at endline |
| Navigation time | Median time on matched route tasks, same participants, baseline versus endline | ≥ 25% reduction, subject to baseline confirmation |
| Route-guidance intervention | Requests for or receipt of directions or physical escort per assessed journey | ≥ 30% reduction from baseline |
| Passenger confidence | Post-route 1–5 confidence item and qualitative interview | +1.0 point mean improvement; distribution reported |
| Women participants | Share of active enrolled participants | ≥ 45%; recruitment gate before full pilot |
| Correct decision-point trigger | Scripted technical walk-throughs across approved routes | ≥ 95%; false advance ≤ 2% |
| Offline availability | Core route instructions available after station-pack download | 100% in acceptance tests |
Budget
USD 500,000 in total.
Ninety per cent requested from ADB. Ten per cent co-financed by E-Sia Learning, valued at production cost.
USD 450,000
ADB grant requested (90%)
USD 50,000
E-Sia co-financing (10%)
USD 500,000
Total project cost
Show the full budget, line by lineHide detail
| Grant budget line | USD | Share | Purpose |
|---|---|---|---|
| Project management, product and instructional design | 82,000 | 18.2% | Delivery leadership, accessible instruction design, reporting and operator coordination |
| Hybrid app, backend and accessibility engineering | 78,000 | 17.3% | Native BLE/PDR client, offline station packs, authoring tools, accessibility hardening |
| Station survey, integration and hardware | 60,000 | 13.3% | Three surveys, anchors and markers, installation, calibration, maintenance stock and training |
| Local DPO coordination and co-design | 70,000 | 15.6% | Chennai coordination, advisory panel, walk-throughs, content review, onboarding and monthly governance |
| Participant honoraria and travel | 25,000 | 5.6% | Time and travel reimbursement for 75 participants and advisory activity |
| Braille-device pool, language and voice | 33,000 | 7.3% | Loan devices plus Tamil, English and Hindi braille, speech and review |
| Independent evaluation and assurance | 52,000 | 11.6% | Evaluation, data protection, security, penetration/offline and accessibility testing |
| Knowledge products and dissemination | 12,000 | 2.7% | Replication blueprint, operator package and accessible dissemination |
| International and in-country travel | 25,000 | 5.6% | Planned technical and governance site visits and local travel |
| Contingency | 13,000 | 2.9% | Currency, site changes and extended acceptance testing |
| Total ADB grant | 450,000 | 100% | Plus USD 50,000 E-Sia co-financing — total project cost USD 500,000 |
Local delivery weighted
DPO coordination, station integration, participant honoraria, the braille loan pool and independent evaluation are all executed in India.
Portable, procurement-ready
Route graphs, anchor IDs and instruction content use documented, exportable formats — no proprietary hardware estate and no guaranteed follow-on award.
Replication envelope
Provisional direct implementation of USD 12,000–20,000 per additional station, to be replaced by verified unit economics after the site surveys.
Principal risks
Named risks, explicit decision rules.
High
CMRL authorisation or site access is delayed
Written operator authorisation is a mobilisation gate. No installation begins without it; any operator substitution requires ADB approval.
High
BLE and sensor estimates produce ambiguous location
Route constraints, calibrated zones, hysteresis and positive anchors; fail safe to stop and re-confirm; no sensor-only platform-edge instruction.
High
A verification marker is hard for a blind user to locate
Consistent placement relative to tactile and handrail cues, user-led discovery tests, and NFC evaluation at selected points.
Medium–high
Crowding and noise reduce usability
Peak and off-peak tests; haptics and braille as private channels; shorter language; closed test before rollout.
Medium
Women are under-recruited or drop out
≥45% recruitment gate, women's networks, reimbursed travel, women-only feedback options and monthly attrition review.
Medium
Partner decisions or handoffs are delayed
Confirmed role matrix, monthly steering group, named workstream owners, decision deadlines and escalation to the project director and CMRL sponsor.
Medium
Privacy or cyber-security weakness
Data minimisation, pseudonymous IDs, consent, encryption, role access, retention schedule and independent testing before full pilot.
Status note: CMRL, Nethrodaya, Prodata Group and CPPR have all agreed the roles presented in this submission. The remaining site, safety, evaluation and production gates are technical approvals held in the pilot's decision log — not unresolved partnership decisions.
Ready for a live demonstration.
We would be glad to show the prototype if shortlisted.
E-Sia Learning · Ghent, Belgium
thuy@e-sia.beSubmission date: 10 September 2026