Pilot
intervention dossier, Katraj–Dehu Bypass, Pune
A worked, decision-ready example of a SpeedTruth Priority-1
corridor: what is wrong, what the evidence shows, what to do, and what
it returns. ADB AI for Safer Roads 2026 · Maharashtra
(India)
1. The corridor
- Where: Katraj–Dehu Road Bypass, western edge of
Pune, Maharashtra (≈18.45–18.69 N, 73.72–73.89 E), a ~52 km arterial
bypass carrying through-traffic around the city.
- What SpeedTruth flags: 10 of its 14
assessed segments are Priority 1 (2 Priority 2); 3 are
absolutely Critical and 7 Serious on
the Safe-System severity scale.
2. The misalignment,
the limit, not the driver
- The corridor is posted at highway speed though it
functions as an urban bypass. Operating speeds run below the
posted limit, drivers are not the problem; the limit itself is
set for a free-flowing highway on a road that no longer behaves like
one, so enforcement alone cannot fix it.
- The road carries dense vulnerable mixed traffic:
auto-rickshaws, motorcycles (helmet-wearing only ~24% in urban
Maharashtra), pedestrians and trucks sharing largely undivided
carriageway with no footpaths. The Safe-System
survivable speed for this mix is 30–50 km/h (WHO), well
below the posted limit. (Specific per-segment posted limits and
probe speeds are confidential under the challenge data agreement and are
omitted here; the classification, diagnosis and recommendation below are
derived outputs.)
- This is a governance-maintenance failure: a bypass
posted at highway speed that urban development and informal traffic have
since grown into, without the limit being revised.
3. The
evidence (this is not inferred, it is photographed)
- Street-level imagery review of four points on the corridor
directly shows the conditions (
IMAGERY_VERIFICATION.md;
images output/blindtest_images/img_01,10,17,21.jpg): an
auto-rickshaw mid-carriageway among motorcycles, a truck and
pedestrians, parked two-wheelers, roadside vendors, dust verges
and no pedestrian footpath, on a road posted at highway
speed.
- These are exactly the “invisible fleet” users
(informal three-wheelers, unhelmeted riders) that probe-speed data
cannot see, which is why the operating-speed signal alone understates
the risk here.
4.
Recommended intervention (aligned with GRSF “what works”)
| Tier |
Action |
Cost |
Evidence |
| Immediate |
Reduce posted limit to 50 km/h (built-up sections)
and 30 km/h at the 5 school/market segments;
re-sign |
Near-zero |
WHO Safe System; a limit re-set is the cheapest credible
measure |
| Medium |
Enforcement + traffic-calming where operating speed still exceeds
the new limit |
Medium |
GRSF |
| Structural |
Separated footways, safe crossings, and motorcycle
lanes where the road must stay fast |
Higher |
GRSF: footways −40–60% pedestrian casualties |
5. What it returns
- The matching iRAP Indian deployments return
benefit–cost ratios of ~13–14 for pedestrian footpaths
and crossings (iRAP India Four States, 2011), the intervention classes
recommended here.
- Because the binding constraint is the posted limit,
the first step is a near-zero-cost quick win:
re-signing the corridor to a survivable speed delivers immediate risk
reduction before any capital works.
- (Honest: with no segment-level crash counts this is a
Safe-System and Power-Model case, not a casualty forecast; see
FINDINGS.md §4, §6.)
6. Why this corridor first
It combines everything ADB prioritises: a
motorcycle/auto-rickshaw corridor with very low helmet
use, school and market exposure, photographed
vulnerable mixing, and a limit that is the lever
(operating speed is already below it, so enforcement alone cannot fix
it). It is a textbook case where correcting the limit, not
chasing the driver, saves lives, and where SpeedTruth turns a
network-wide scan into a single, costed, actionable decision.
Method and validation: METHODOLOGY.md. Evidence:
IMAGERY_VERIFICATION.md. Sources:
SOURCES.md.