See small drones before they cross the border, and close the gaps they fly through.

Drones now carry most cross-border drug deliveries into Punjab. BSF recovered 21 drones on the Punjab border in 2022, 107 in 2023 and 294 in 2024 (The Tribune, 14 Nov 2025). Of 305 drone-linked drug trafficking cases recorded in India in 2025, 298 were on the Punjab border (Ministry of Home Affairs reply in the Rajya Sabha, via Millennium Post, 22 Jul 2026).

The flights are hard to stop. They peak between 22:00 and 04:00, drop in fields 3.5 to 10 km inside, often follow pre-programmed GPS routes that RF detectors cannot hear, and fly home when jammed (Indian Express via SuperKalam, 12 Sep 2026). Own drones carry their own risk: on 5 May 2026 a BSF surveillance drone came down 10 to 20 yards inside Bangladesh at Patgram (Prothom Alo).

• Shows every official drone figure with who counted it, the period and the report, checks each report still says it, and adds a sample of individual incidents from the news, with no person named.
• Plans sensor coverage on the 238 km Amritsar and Tarn Taran line: radar, EO/IR, RF and acoustic, with line and drop-corridor coverage, gaps, warning time and the fewest sensors that close them.
• Maps advisory drone zones around 77 border-region aerodromes and the 25 km border belt, and sizes an own-drone geofence from speed, wind, reaction time and navigation error.
• Replays a simulated night with tracks, classification confidence, alerts and human-approved responses, compares layouts over 50 simulated nights, and exports the night as a trial log.
• Gives a sourced response playbook: detect, classify, alert, respond lawfully through authorised agencies, keep evidence, recover payloads safely, and test equipment before fielding it.

1. DetectLayered radar, cameras, RF and acoustic sensors along the line and in depth.
2. ClassifyEach cue changes the odds a track is a drone; a camera or RF match confirms it.
3. AlertWatch, warning, danger and search levels, each with a recommended action and a police handoff.
4. RespondOnly authorised agencies, only lawful non-kinetic means, and a person decides.
5. EvidenceLogs, video and the drone itself kept under a chain of custody.

Every formula, threshold and assumption is in the methods note and on the console's Methods page, with its source.

• Border units and sector headquarters: where the line and the drop corridor lack coverage, and what closes the gap.
• State police anti-drone and checkpoint units: when and where drops happen, and the response sequence that preserves evidence.
• Own-drone operators: a geofence setback that holds in wind, after a link loss and under GNSS interference.
• Airport and airspace authorities: which aerodromes sit near the border, and where counter-drone measures need coordination.

• Defensive by design: detection, tracking, classification, alerting, lawful non-kinetic response planning and evidence. No weapon-targeting guidance.
• Responses only through authorised agencies: counter-drone jammers need Ministry of Home Affairs permission (Ministry of Defence for the defence forces) under the Cabinet Secretariat's jammer guidelines.
• Software recommends; trained people decide. Every response, including "no action", is logged.
• Events and places, never people: headlines are not stored, arrest stories are excluded, and no person, border outpost or troop position is mapped. Aerodromes come only from public aeronautical data (OurAirports), including air force stations, because own drones must keep clear of them.
• Simulated data is marked as simulated on every page and in its JSON. Airspace zones are advisory; the Digital Sky map is the authority.

• Sensor ranges are planning assumptions. Tree lines, crops, villages, fog and RF noise change them; plans need a site survey.
• The border line is approximate and must not be used for navigation or to program a geofence.
• Official figures come through press reports, some second-hand; agencies count differently.
• The news sample is incomplete by nature and is never presented as a count.
• The simulator's likelihood ratios and response effects are assumptions, not field measurements.

One sector for one season. We load your sensor list and ground survey, measure real detection ranges against these plans, replay your own logs through the classifier, and produce trial logs in the form a certifying committee can review. contact@trinetra.in