Skip to main content

Development

Pollution Control Board uses UNDP's GeoAI satellite platform to find polluting brick kilns

In February 2021 the Bihar State Pollution Control Board signed an MoU with UNDP India to use GeoAI, a platform built by UNDP's Accelerator Lab and the University of Nottingham that finds brick kilns in satellite images with an AI object-detection model and flags those likely to be non-compliant. Inspectors then check the flagged kilns with a mobile app. A 2023 report prepared with the board records field inspection of 1,680 kilns in five districts. The board has not published statewide use or enforcement results.

AIGovernmentBihar, India
Stage
Programme or pilot initiated
Confidence
High
Latest news
Sources
5 cited

Journey so far

The highlighted step is the current stage. Not every development passes through every step.

  1. Announcement only
  2. Agreement approved
  3. Agreement finalised
  4. Programme or pilot initiated (current stage)
  5. Operational
  6. Outcome verified

Good to know: An MoU followed by GeoAI-guided inspections documented in five districts. Outcome figures come only from UNDP, so this is recorded as a pilot rather than a statewide operation or a verified outcome.

45%
Details confirmed so far5 of 11 tracked details

What's confirmed

5 facts backed by sources

  • The board signed an MoU with UNDP India, signed online by board chairman Dr Ashok Kumar Ghosh and UNDP's Shoko Noda and reported on 10 February 2021, to use the GeoAI platform developed by UNDP's Accelerator Lab and the University of Nottingham.
  • The board's stated plan was to begin with brick kilns and later use the platform for stubble-burning sites and vehicle-pollution hotspots in all 38 districts.
  • UNDP said in September 2022 that GeoAI had mapped more than 7,800 kilns in Bihar, prioritised 1,000 and that more than 600 had been inspected and acted on by regulators.
  • The 2023 report, prepared by the board with UNDP, Development Alternatives and the University of Nottingham, records field inspection of 1,680 kilns across five districts. It reports that 96% of detected kilns were found at the detected location and that the model classified zigzag versus fixed-chimney kilns correctly about 65% of the time.
  • The same report sets 31 December 2023 as the deadline for kiln registration.

Still to be confirmed

6 things to watch for

  • Any use of GeoAI for stubble burning or vehicle pollution. Nothing after the 2021 announcement mentions either.
  • Enforcement outcomes. UNDP's statement that more than 600 kilns were 'acted upon' is a partner's claim; the board has not published how many kilns were closed, converted or penalised because of GeoAI targeting.
  • Kiln totals differ: about 6,000 (board chairman to TOI, 2021), 7,500 with almost 4,000 converted (chairman quoted by UNDP, 2022) and more than 7,800 mapped (UNDP, 2022). The unconverted share also differs for the same 2021 statement: about 1,000 of 6,000 (TOI) against 30% of 6,000 (IndiaAI). All are recorded and none is preferred.
  • Whether the board still uses GeoAI after December 2023, and whether use went beyond the five districts.
  • The MoU text, its term and any cost to the state.
  • UNDP Accelerator Labs case-study figures (1,000 kilns transitioned, 500,000 tonnes of CO2 avoided a year) could not be opened or dated and are not recorded.

How it unfolded (3)

Each step as a dated source records it, oldest first

  1. MoU approved

    Board chairman Ashok Kumar Ghosh and UNDP India's Shoko Noda signed an MoU online for the board to use the GeoAI platform, starting with brick kilns.

    The Times of India · IndiaAI (Ministry of Electronics and IT)

  2. Pilot

    UNDP said GeoAI had mapped more than 7,800 kilns in Bihar and prioritised 1,000 by non-compliance score, of which regulators had inspected and acted on more than 600.

    UNDP India

  3. Pilot

    UNDP published 'GeoAI for Brick Kilns in Bihar', a report prepared with the board, recording field inspection of 1,680 kilns across five districts with the GeoAI app.

    UNDP India (publication page) · BSPCB, UNDP, University of Nottingham and Development Alternatives, 'GeoAI for Brick Kilns in Bihar' (report PDF, (c) UNDP 2023, hosted by Development Alternatives)

Sources (5)

Who reported it, when, and what they cover

PublisherPublishedCoversLink

The Times of India

Established news

  • • MoU signed 'on Tuesday' between the board and UNDP
  • • Signatories
  • • Plan for kilns, stubble burning and vehicle hotspots in 38 districts
  • • About 1,000 of 6,000 kilns not yet on zigzag technology
Open source (opens in a new tab)

IndiaAI (Ministry of Electronics and IT)

Official government

  • • MoU signed virtually by Ghosh and Noda for access to the GeoAI platform
  • • Chairman: of 6,000 kilns, 70% had adopted cleaner technology
  • • Page served with an expired TLS certificate when read
Open source (opens in a new tab)

UNDP India

International organisation

  • • More than 7,800 kilns mapped in Bihar; 1,000 prioritised; more than 600 inspected and acted on
  • • Board chairman: almost 4,000 of 7,500 kilns switched to greener technology
  • • undp.org returns 403 to scripts; read through the Wayback Machine
Open source (opens in a new tab)

UNDP India (publication page)

International organisation

  • • Report developed in partnership with the Government of Bihar, Development Alternatives, the University of Nottingham and UNDP
  • • Publication date 20 December 2023 (read through the Wayback Machine)
Open source (opens in a new tab)

BSPCB, UNDP, University of Nottingham and Development Alternatives, 'GeoAI for Brick Kilns in Bihar' (report PDF, (c) UNDP 2023, hosted by Development Alternatives)

International organisation

  • • Report prepared by the Bihar State Pollution Control Board with partners
  • • 1,680 kilns inspected across five districts
  • • 96% of detected kilns found; about 65% correctly classified
  • • Kiln registration deadline of 31 December 2023
  • • The document itself is dated 2023; the day comes from the UNDP publication page
Open source (opens in a new tab)

Record ID: BIHAR-AI-2021-02-001Dataset last checked Spotted an error? Report a correctionHow corrections work