Road-environment audit · SSI v0.3

The road outsideRaghubir Singh modern school

126, Road No. 7, Rajan Enclave, Vikas Nagar, Deepak Vihar, Uttam Nagar, Delhi, 110059, India28.63381, 77.04717audited 2026-08-06

15/100

Critically unsafeprovisional band

Critically unsafe. Almost nothing on this road protects a child from traffic on the way to the gate.

Of the five things we check, 3 are missing entirely.

This road · 15

0 · no protectioncomplete protection · 100

Checked twice, independently

Two AI readers on separate accounts · corrections logged in public

Confirmed

Corrected

Unresolved

See how each was judged

Of 153 roads scored so far, this one ranks 105 (worst first). The others scored between 0 and 41.

11.25points earned

76points we could see (76%)

14.8%provision

1exposure — local

14.8provision × exposure

→

15SSI · rounded to nearest · v0.3

Only what the imagery could actually resolve is scored (76% of the rubric here); the rest is left uncounted rather than guessed. A road with full protective infrastructure scores near 100.

The road in context · derived from public data

Where this road sits

Before the checks: what kind of road this is, how it meets other roads, and what passes it. These lines describe the road and its surroundings — they do not change the score.

Road layout around the gate, drawn from OpenStreetMap — Road layout · drawn from OpenStreetMap · blue dot = school coordinate

Satellite view of the same area — Satellite · © Google · retrieved 2026-08-13 (capture date not published by the imagery API)

FrontageThe gate opens onto a local street, on its E side

Junctionthe gate area sits at a 4-leg junction

Street activityno bus stop or shops mapped within 150 m (OpenStreetMap)

Crash recordcrash-location data is not yet available for Delhi — this line will be added when a public source lands

Exposure contextModerate — 22/100 (partial: no crash component yet). Describes how much traffic and crash risk passes this road — context only, never part of the road score (method: exposure v1, Data page)

Road layout data © OpenStreetMap contributors (ODbL), retrieved 13 Aug 2026 · crash data: Jaipur Police · derivations: `exposure_v1` on the Data page.

What we found · five checks · IRC:SP:32-2023

Five things we checked — and the photo behind each

Each check was read once, then verified blind by a second reader on a separate account. The photo each finding came from sits right beside it. Tap “see how this was judged” for the full reasoning.

Slowing traffic downP1 · Physical speed control

Nothing slows traffic at the gate

No speed hump, table or rumble strip on the approach — vehicles reach the gate at full speed.

Absent

Also cited for this check

0 of 30 points

LevelNone (L0)

Points0 of 30

ConfidenceHigh

Coverage seen≈80%

VerificationConfirmed — both readers agreed

The full reasoning

No speed hump, speed table, raised crossing, rumble strip or transverse bar marking visible anywhere on the frontage or on either approach. Road surface is uniform unmarked asphalt/concrete in all frontage and traverse views, including both approaches within 30 m of the gate. Coverage is good from ~30 m west to ~46 m east of the gate; the far ends of the zone are only seen at low resolution through haze, and the immediate camera foreground is masked in every frame, so a small worn device directly under a camera position cannot be fully excluded — but overlapping mid-ground from four frontage panos covers those gaps.

Getting across the roadP2 · Safe crossing

No safe place to cross

No marked or raised pedestrian crossing at or near the gate.

Absent

Also cited for this check

0 of 25 points

LevelNone (L0)

Points0 of 25

ConfidenceHigh

Coverage seen≈80%

VerificationConfirmed — both readers agreed

The full reasoning

No marked pedestrian crossing, no faded remnant of one, and no refuge island anywhere in the imagery. The carriageway carries no markings of any kind — no zebra, stop line, edge line or centre line — over the whole frontage and both approaches, including the surface directly outside the gate. Undivided carriageway, so no refuge requirement is triggered. Same coverage limits as P1 at the far ends of the zone.

◑

Walking without stepping into trafficP3 · Footpath & separation

A broken walking edge

A walking edge exists but is broken or blocked for most of the frontage.

Partial⟳ Corrected in review

NoneMinimal

An independent second reader changed this — the two readings disagreed, and the evidence below settled it.

Also cited for this check

6.25 of 25 points

LevelMinimal (L1)

Points6.25 of 25

ConfidenceMed

Coverage seen≈70%

VerificationCorrected in the second reading

The full reasoning

The audit read (L0) falls. Decided on ZJJnovrEGdbH_h271_f90_road.jpg and 9Q0R7ks6_BVf_h090_f90_trav.jpg, re-examined at stage 2, which frame the school-side frontage from opposite directions. Both show a strip of appreciable width between the compound wall/railing and the traffic lane, at carriageway level with no resolvable kerb, broken by the gate lodge building line meeting the road edge and by planted poles; the north side has fragmented private plinths at varying levels. That is the L1 description — a walkable edge that exists but is broken by permanent causes along most of the frontage — not L0 'none'. The audit's L0 rests on two bases that the methodology disallows or that do not apply: pedestrians and children being on the carriageway (behavioural, and the parameter is explicitly permanent built condition only), and Zg0Xh0oyy4f5_h005_f90_trav.jpg cited for 'zero shoulder', which is a separate residential lane ~150 m SW, not the school frontage. Stripping those, the audit's own surviving observations ('walls abut the road directly in several segments', 'aprons fragmented by poles/steps') describe a broken edge rather than no edge. The blind read was also attacked: its 'paved strip' is unpaved dust in stretches and is vehicle-occupied in every view — but vehicle occupation is momentary and must be ignored here, so it cannot push the level down. Resolved rather than left unresolved because the disagreement is definitional (does a flush discontinuous edge count) and not a limit of the imagery. Confidence MED, not HIGH: no kerb line or delineation is resolvable and the edge surface is occluded in parts of every frontage view. Observability settled at 0.7 (reads differed by 0.05); the gap to 1.0 is the frontage edge surface hidden under standing vehicles and stalls plus the masked near-camera strip in each frame. L1 does not require HIGH confidence, so MED does not force a downgrade.

Being seen and expectedP4 · Signage & zone identity

No school-zone signs

Drivers get no warning that a school gate is here.

Absent

Also cited for this check

0 of 10 points

LevelNone (L0)

Points0 of 10

ConfidenceMed

Coverage seen≈70%

VerificationConfirmed — both readers agreed

The full reasoning

No school warning sign on either approach, no speed-limit sign or carriageway speed stamp, and no 2023-identity asset (no transition or zone-start sign, no fluorescent yellow-green sign faces, no red-white school-zone carriageway markings, no beacon hardware). The only roadside signs at the gate are two small white parking-restriction plates mounted on the gate leaf (ZJJnovrEGdbH_h207_f35_gate_zoom.jpg) and a compound-mounted notice board (2S_H4m33hjEC_h239_f35_gate_zoom.jpg); neither is a school warning sign. Near-gate view is clear; confidence held at MED because signs further out on the approaches would be small and are not resolvable in the haze, and the roadside is dense with poles, cables and commercial hoardings.

◑

The gate itselfP5 · Gate environment

Space and a railing at the gate

A clear setback or railing gives room to gather off the carriageway.

Partial

Also cited for this check

5 of 10 points

LevelBasic (L2)

Points5 of 10

ConfidenceMed

Coverage seen≈75%

VerificationConfirmed — both readers agreed

The full reasoning

Built form: the gate leaves are recessed behind a paved forecourt rather than opening directly onto the carriageway, and a metal railing on a low wall runs along the compound frontage, giving off-carriageway standing space between the railing line and the road edge. Setback depth is consistent across both gate panos. Not L3: no separate pedestrian wicket is confirmable in the zoom (the brick lodge blocks part of the gate opening), and there is no crossing for a rail to run to. No marked pick-up/drop-off bay visible anywhere on the frontage. Confidence MED — setback is evident in two views but its depth and the gate's full width are partly occluded by the lodge and the foreground camera mask.

At the gate

On the frontage road

Along the street

Some captures sit on neighbouring lanes; older ones are shown greyed and were used only for context.

What must be fixed · 5 actions

What would make this road safe — and who owns each fix

Each action maps to a failing check, the relevant IRC:SP:32-2023 clause, and the responsible agency. Agency labels are provisional and correctable through the dispute channel.

01Lifts P1

Install a speed table or twin speed humps within 30 m of the gate

IRC:SP:32-2023 §5.13 · Table 5.1MCD / PWD · provisional

Speed humps near schools roughly halve child-pedestrian injury odds (Tester 2004); 30 km/h calmed zones ≈ 40% fewer casualties (Grundy 2009).

02Lifts P2

Mark a crossing within 20 m of the gate — table-top preferred

IRC:SP:32-2023 §5.4.3MCD / PWD · provisional

Raised crossings cut vehicle speeds at the exact conflict point.

03Lifts P3

Provide a continuous clear walking edge on the school side

IRC:SP:32-2023 §5.4.1MCD / PWD · provisional

A continuous footpath keeps children off the carriageway on the approach, not only at the gate.

04Lifts P4

School-zone warning signs on both approaches with the speed limit

IRC:SP:32-2023 §5.8MCD / PWD · provisional

The zone must announce itself to drivers before the gate.

05Lifts P5

Add space and a marked pick-up bay off the carriageway at the gate

IRC:SP:32-2023 §5.2MCD / PWD · provisional

Children should not step from the gate into a moving traffic lane.

Responsible-agency labels come from the CFI field instrument’s mapping and are correctable via the dispute channel.

Re-audit timeline

2026-08-06Audited

2027-02-06Re-audit due

Every published road is re-read every six months, or as soon as new Street View imagery lands, or on a sustained dispute.

Something wrong on this page?

The entrance, the imagery, a finding — if any of it is wrong, tell us. A dispute pauses this page while the audit is re-verified, and the correction is logged in the public record.

Dispute this audit