Using AI/ML and Space Technology to Identify Manganese Reserves and Overcome Production Shortfalls.
Ministry of Steel · Smart Automation · Software
The satellite inputs this statement names cannot identify manganese and no operational data is supplied, so unless you are willing to quietly rebuild the reserve half around spectral geology and be candid about it, you are building a dashboard over an empty foundation.
What it actually is
MOIL estimates how much manganese ore it has using manual surveys and drilling, which is slow and often wrong, so actual production misses the plan. The ask is a system that uses satellite data and past production records to map reserves better and warn when output is about to fall short. It should also suggest what to change to fix the shortfall.
What to build
Two connected tools behind one dashboard: a prospectivity layer that scores areas for manganese potential from remote sensing inputs and existing geological maps, presented as a map with confidence bands rather than a hard reserve figure; and a production shortfall forecaster taking historical output alongside equipment availability, downtime, rainfall and blasting schedule to project the coming period's tonnage against target, flag the specific constraint driving any gap, and generate the corrective recommendations the description names — schedule adjustment, blasting optimisation, equipment redeployment — with the projected recovery from each.
Smallest thing that wins the room
Show the shortfall forecaster catching a projected production gap several weeks out, attribute it to a specific constraint, and show the tonnage recovering when that constraint is relaxed in the model.
How crowded this one gets
A guess, projected from the 2025 statements — the last year where both the submission counts and the winners were published.
Quieter than 58% of the 226 · #95 of 226 by expected field
A normal-sized field. Your idea has to be good, not miraculous.
Why: central ministry statements sat below the average.
This is a guess, not a fact
Nobody has published 2026’s numbers yet. This is an analysed estimate from last year’s pattern, so please do not take it as the truth — check the live counter on the SIH portal before you decide anything. The range covers the middle half of likely outcomes, so one statement in two lands outside it. Entry closes at 500 ideas per statement, so no range goes past that — a statement that reaches the cap fills and shuts rather than drawing an unlimited crowd. The model reads only three things a team can see before choosing — software or hardware, the theme, and what kind of body posted it — and those explain about a quarter of the variation in last year’s field sizes (R² 0.25 on held-out statements). Trust the band more than the number, and the ordering more than either. It cannot see how good your idea is, which is the part that actually decides it.
The scores
The number is the shorthand. The line under it is the reason.
Acceptance potential
2/5The reserve identification half is built on a hollow premise that a geologist judge will dismantle immediately, and the shortfall half depends entirely on internal operational data you have no route to, which leaves a well-built dashboard sitting on top of nothing verifiable.
Feasibility
2/5Sub-surface manganese reserves cannot be identified from the satellite inputs the description names — rainfall, soil moisture, NDVI and land surface temperature are not mineralisation indicators — and the data that would work, ASTER spectral or airborne geophysics plus MOIL's own drill logs, is either unavailable or internal, while the production and equipment downtime records the second half needs are proprietary and will not be released to a student team.
Innovation scope
4/5Only a dashboard is prescribed and the description is short enough that the modelling approach, the choice of indicators and the entire structure of the recommendation engine are left to you.
Clarity
2/5At barely 1,300 characters it fuses two unrelated problems — geological prospecting and operations scheduling — into one statement, and the satellite inputs it names have no established causal link to the reserve estimation it is asking for, so the premise itself is confused rather than merely thin.
Effort
HeavyTwo independent modelling problems in different disciplines, each needing its own data pipeline and validation story, plus a unified dashboard — the code is not deep but the ground you have to cover is wide and the two halves share nothing.
Demo-ability
HardThere is no ground truth in the room for either half — you cannot show a predicted reserve is real without drilling it, and you cannot show a production forecast is right without MOIL's actual subsequent output.
In its favour
- Green flag: Innovation scope is wide open because the description prescribes almost nothing beyond a dashboard, so you can define the problem in a way you can actually deliver
- Green flag: Mineral exploration is an unusual domain for a hackathon and the field for this statement will be very thin
- Green flag: The production forecasting half is a legitimate, well-understood time-series problem that works on synthetic operational data if you frame it as a planning tool rather than a prediction of MOIL's real output
- Green flag: ASTER band-ratio indices for manganese-bearing formations are documented in published remote sensing literature, giving you a defensible method if you take the spectral route rather than the one the description suggests
Against it
- Red flag: The description asks you to find sub-surface reserves using rainfall, soil moisture, vegetation index and land temperature, and none of those indicate manganese — following the statement literally produces a model that is scientifically indefensible
- Red flag: MOIL's geological drill logs, production records and equipment downtime data are all internal, and no dataset link is provided, so both halves start with a data source you have to invent
- Red flag: Two unrelated problems in one statement means a team either splits its effort and does both badly or answers half the PS
- Red flag: There is no way to validate a reserve prediction without drilling, so your headline claim is permanently unverifiable and a domain judge knows that better than you do
What you will be writing
- ASTER / Sentinel-2 spectral band ratio mineral indices
- Google Earth Engine raster processing
- Random Forest prospectivity mapping
- SARIMA production time-series forecasting
- GeoServer + Leaflet map dashboard
- scikit-learn constraint attribution
- Mineral exploration
- Mine production planning
- Remote sensing
Prior art to read before you start
mineral prospectivity mapping · production shortfall forecasting · satellite-derived geological indicators
Analysed by Claude Opus. Every score above is a judgment call with its reasoning attached — kindly cross-check this against the official statement on the SIH portal before your team commits to it.