Skip to content
MargRakshak AI — Roadside Hazard Detection NetworkPre-pilot · Corridor New DelhiSpec v3.0 · Build 2026.11
MRAI logo
MRAI
by UniWebs
All clear
14ms
Document register

Known Limitations & Non-Detection Cases

What the network does not detect, where it degrades, and what a human still has to do.

Document
MRAI-DOC-006
Revision
Rev. 1.2
Effective
2026-08-27
Classification
Public

01 · Not detected

Events outside the node's field of view or beyond ultrasonic range.

Slow-developing hazards: potholes forming, structural damage, ice.

Anything requiring identification of a vehicle or person — deliberately out of scope.

02 · Degraded conditions

Heavy rain, fog, and glare reduce vision confidence. Night performance depends on ambient street lighting.

Gas readings drift with temperature and humidity and are treated as indicative, not quantitative.

Ultrasonic ranging is unreliable against soft or angled surfaces.

03 · Human in the loop

Every critical incident is confirmed by an operator before any external action.

The network reports conditions. It does not control signals, barriers, or traffic.

Issued by UniWebs — Kavy (software) & Saksham (hardware)
Operator of record
UniWebs — Kavy (software) & Saksham (hardware)
Independent student-run studio · not a government body
Enquiries
operations@projectmrai.in
Revision
Spec v3.0 · Build 2026.11
Last reviewed 2026-11-02