Precision agriculture and drones

Agricultural Engineering (B.E./B.Tech) · Structures and technology · free study notes for exams, placements and interviews

In short

Precision agriculture is the practice of measuring variability within a field and managing each part of it according to its own need — the right input, in the right amount, at the right place and time — instead of treating the whole field as uniform. It is built on three legs: sensing (GPS/GNSS, soil sensors, imagery, yield monitors), decision-making (management zones, prescription maps, decision support software) and actuation (variable-rate applicators, auto-steer tractors, spray drones). In India it has moved from research to policy: the Kisan Drone scheme, SMAM/AMFU subsidies, Namo Drone Didi, drone-spray SOPs approved by CIB&RC and the DGCA Drone Rules, 2021 with the Digital Sky platform. It is examined in Agricultural Engineering and Farm Machinery papers, in ICAR-JRF/SRF and GATE (AG) under farm power and machinery plus remote sensing, and asked at interviews by Mahindra, John Deere, Escorts-Kubota, TAFE, Dhanuka, Coromandel, IFFCO and drone firms like Garuda Aerospace, IoTechWorld and Marut Drones.

Explained

Concept, scope and the precision-farming cycle

Definition: site-specific crop management (SSCM) — matching inputs to spatially and temporally variable crop needs, using geo-referenced data.
Sources of variability: soil texture and depth, organic carbon, pH/EC, nutrient status, topography and drainage, pest/weed patches, past management, irrigation uniformity.
The closed loop (examinable as a cycle): (1) data acquisition → (2) data processing/mapping → (3) interpretation and decision → (4) differential application → (5) yield/record monitoring → feeds back into (1).
Enabling technologies: GNSS positioning, GIS, remote sensing, proximal sensors, variable-rate technology (VRT), telematics, machine learning, drones/UAVs, IoT.
Two implementation routes: *map-based* (survey → map → prescription map → apply with GNSS position look-up) and *sensor-based / real-time* (on-the-go sensor reads crop or soil and the controller acts in the same pass, no map needed). Hybrid systems use a map plus real-time correction.
Economic logic: worth doing only when (value of yield gain + input saved) > (cost of sensing + equipment + skill). Hence India's adoption begins with drones and service models (custom hiring) rather than farmer-owned machinery.

Positioning, data capture and mapping

GNSS: GPS (USA), GLONASS, Galileo, BeiDou and India's NavIC/IRNSS (ISRO, 7-satellite regional constellation; L5 and S-band). Plain GNSS gives metre-level accuracy.
Differential correction: DGPS uses a reference station to remove common errors → sub-metre. RTK (real-time kinematic) uses carrier-phase corrections over radio/NTRIP → ±2–3 cm, needed for auto-steer, bed-planting and inter-row weeding. SBAS/GAGAN is India's satellite-based augmentation for aviation-grade correction.
Grid soil sampling: divide the field into cells (commonly 0.4–1 ha; finer for high-value crops), take 8–15 cores per cell, composite, geo-tag, analyse, then interpolate.
Interpolation: inverse distance weighting (IDW) and kriging (uses the semivariogram — nugget, sill, range) produce continuous nutrient/property surfaces.
Management zones: cluster layers (EC, elevation, yield history, NDVI) into 2–5 zones; zone sampling is far cheaper than dense grids and is the practical Indian route.
Proximal sensors: soil EC (Veris, EM38) as a texture proxy, capacitance/tensiometer soil-moisture probes, optical crop sensors (GreenSeeker, Crop Circle) measuring NDVI on the go, chlorophyll meter/SPAD and leaf colour chart (ICAR-promoted low-cost N tools).
Yield monitoring: combine-mounted impact or optical flow sensor + moisture sensor + GNSS + ground-speed sensor + header-width → georeferenced yield map; it is the scorecard of the whole system.
Remote-sensing indices used for decisions: NDVI (vigour), NDRE (N in dense canopy), NDWI/NDMI (water), SAVI (sparse canopy, soil correction), thermal (canopy temperature → CWSI for irrigation scheduling).

Variable-rate technology and guidance

VRT components: controller/display, GNSS receiver, prescription map (shapefile/ISOXML), flow or metering control (hydraulic drive, servo motor, PWM solenoid), and feedback sensor; ISOBUS (ISO 11783) is the machine-electronics standard, with the ISO 11783-10 task file for maps.
VRT applications: fertiliser (spinner-disc or pneumatic spreaders with variable belt speed), seed rate and depth, pesticide (PWM nozzles, section control), lime/gypsum, and irrigation (VRI pivots, zoned drip).
Guidance levels: (1) manual lightbar, (2) assisted steer, (3) auto-steer with RTK, (4) controlled-traffic farming, (5) autonomous/driverless tractors — Mahindra and John Deere already sell auto-steer and telematics in India.
Section and swath control: shuts nozzles or rows off in overlaps, headlands and point rows; typically saves 4–10% of input on irregular fields.
Sprayer electronics: pressure-based rate control vs PWM (pulse-width modulation) which holds droplet size while varying flow; direct injection keeps chemical out of the tank until needed.
Precision irrigation: soil-moisture-sensor or ET-based automated drip/fertigation, flow and pressure logging; links to PMKSY "per drop more crop" funding.

Drones (UAVs/RPAS) in agriculture

Classification by DGCA weight class: nano ≤250 g; micro >250 g–2 kg; small >2–25 kg; medium >25–150 kg; large >150 kg. Most Indian agriculture spray drones are "small" (10 L tank, ~25 kg AUW) and survey drones are micro/small.
Configuration types: multirotor (quad/hexa/octo — VTOL, low speed, high payload-per-size, short endurance, best for spraying), fixed-wing (long endurance, large area mapping, needs runway/catapult), hybrid VTOL, and single-rotor helicopter.
Payload/sensor types: RGB camera (visual, photogrammetry), multispectral (4–10 bands: green, red, red-edge, NIR → NDVI/NDRE), hyperspectral (hundreds of narrow bands, research), thermal (canopy temperature, irrigation, livestock), LiDAR (canopy height, terrain under canopy), and spray/granule payloads.
Agricultural uses: crop health and stress mapping, nutrient status, pest/disease scouting, weed patch mapping, plant counting and gap detection, crop-cut and acreage estimation for insurance (PMFBY), damage assessment after storms/floods, orthophoto and DEM for land levelling and drainage design, spraying and granule/seed broadcasting, livestock and asset monitoring.
Mapping workflow: mission planning (altitude, 70–80% front and 60–70% side overlap) → autonomous flight with GSD target → image capture with GNSS/IMU tags → radiometric calibration (reflectance panel/sunlight sensor) → SfM photogrammetry → orthomosaic + DSM → index raster → zone/prescription map.
Spraying advantages: 1 acre in roughly 6–10 minutes against 1.5–3 hours by knapsack; 90% less water (ultra-low-volume, 10–25 L/acre vs 150–250 L/acre); no operator exposure or trampling; can fly over tall, standing or waterlogged crops.
Spraying limitations and physics: small droplets drift; rotor downwash aids canopy penetration but also carries spray; swath width 3–5 m depends on height (2–3 m above canopy) and rotor diameter; VMD of 100–300 µm typical; efficacy depends on droplet density (droplets/cm²), not just litres; systemic and translaminar chemistries suit low volumes better than contact fungicides.
Nozzles and atomisers: hollow-cone and flat-fan hydraulic nozzles, rotary atomisers (controllable droplet size by disc rpm), and electrostatic nozzles that charge droplets for better underleaf deposition.
Other hardware: brushless motors and ESCs, LiPo/Li-ion smart batteries (10–20 min flight), RTK module, obstacle-avoidance radar, terrain-following radar, flow meter and level sensor, foldable arms, flight controller with geofence and return-to-home.

Indian regulation, standards and schemes

Drone Rules, 2021 (Ministry of Civil Aviation), with later amendments, replaced UAS Rules 2021: abolished many approvals, introduced the Digital Sky platform, UIN registration, Type Certification by DGCA, and the colour-coded airspace map (green zone — no permission below 400 ft; yellow — ATC permission; red — prohibited).
Remote Pilot Certificate (RPC) issued through a DGCA-approved Remote Pilot Training Organisation (RPTO); required for small/medium class drones (nano and micro-for-non-commercial are exempt). Minimum age 18, maximum 65, Class-10 pass.
Certification Scheme for Unmanned Aircraft Systems (CSUAS) sets the type-certification requirements; manufacturers must also meet the no-permission-no-takeoff and real-time-tracking/ID requirements as notified.
Pesticide application: the Insecticides Act, 1968 and Insecticides Rules, 1971 govern use; CIB&RC has approved label claims for drone application for a growing list of formulations, and the SOP for drone application of pesticides (2022, Ministry of Agriculture & Farmers Welfare) fixes operational rules — buffer distances from habitation and water bodies (commonly 10 m type restrictions), wind-speed limit (spraying generally avoided above about 3 m/s / 10 km/h), PPE, no spraying in high temperature or inversion conditions, and a trained operator plus record-keeping. *Approved chemical lists and exact buffer figures are revised periodically — always check the current CIB&RC/SOP notification.*
Schemes and support: Kisan Drone promotion under SMAM — subsidy historically up to 100% (₹10 lakh cap) for ICAR/KVK/State Agricultural Universities, 75% for FPOs, 40–50% (₹4–5 lakh) for individual farmers in notified categories, and 40% (₹4 lakh) for Custom Hiring Centres; contingency support per hectare for drone hire. Namo Drone Didi provides drones to women's SHGs with central assistance. Drone Shakti, PLI for drones and components, and the ban on import of foreign-made drones (components exempt) push domestic manufacture. *Subsidy percentages and caps change with each year's guidelines.*
Data and spatial policy: the National Geospatial Policy, 2022 and the 2021 geospatial data guidelines liberalised Indian map data for domestic entities; Agristack / AgriStack-based Digital Agriculture Mission, Krishi-DSS, farmer and crop-sown registries and geo-referenced village maps form the national data backbone.
Other relevant codes: BIS standards exist for sprayers and nozzles (e.g. knapsack/power-sprayer specifications) and IS 10064 covers spraying/dusting terminology; drone-specific BIS standards are still evolving, so quote the DGCA CSUAS requirements rather than an IS number.

Advantages, limitations and Indian constraints

Advantages: 10–30% fertiliser saving and 5–15% yield gain reported in Indian trials (varies widely by crop and zone); 80–95% water saving in spraying; drastically lower labour and operator exposure; better traceability and insurance claim settlement; reduced run-off and eutrophication.
Technical limitations: high capital cost (spray drone ₹5–10 lakh, mapping drone ₹2–6 lakh, RTK-auto-steer kit ₹3–8 lakh); short battery endurance; cloud cover limits optical satellite data (drones solve this); calibration and georeferencing skill; interoperability of file formats.
Indian structural constraints: average operational holding about 1.08 ha (Agriculture Census 2015-16) and highly fragmented plots, so within-field variability may be smaller than between-plot variability; poor rural connectivity for RTK/NTRIP; shortage of trained remote pilots and agronomic interpreters; no agreed prescription algorithms for many Indian crops.
The answer to small holdings: service delivery — Custom Hiring Centres, FPOs, drone-as-a-service (₹300–600 per acre typical spraying charge), village-level entrepreneurs, and "block-level" rather than "within-field" precision.
Comparison of sensing platforms: satellite (free/cheap, 5–30 m, cloud-limited, frequent), drone (cm-level, on demand, small area, costlier per ha for large areas), ground sensor (point accurate, no spatial cover), manned aircraft (rare in India).

Key points

Precision agriculture - managing within-field variability with geo-referenced data so each zone gets its own input rate.
Four Rs - right input, right rate, right place, right time; the operating definition examiners accept.
Map-based vs sensor-based VRT - survey-and-prescription-map versus real-time on-the-go sensing and actuation in one pass.
RTK - carrier-phase differential GNSS giving ±2–3 cm, needed for auto-steer and inter-row operations; DGPS gives sub-metre.
NavIC/IRNSS - India's regional navigation system; GAGAN is the SBAS augmentation.
Kriging - geostatistical interpolation using the semivariogram (nugget, sill, range) to build soil property maps.
Management zones - 2–5 clustered sub-areas; cheaper and more practical than dense grid sampling in Indian conditions.
Yield map - georeferenced output from a combine's flow + moisture + GNSS + speed sensors; the feedback leg of the loop.
NDVI = (NIR − Red)/(NIR + Red) - vigour index; NDRE replaces red with red-edge for dense canopies to avoid saturation.
GSD - ground sampling distance (cm/pixel); set by sensor pixel size, focal length and flight altitude.
Drone weight classes - nano, micro, small, medium, large under the Drone Rules, 2021; agricultural sprayers are "small".
Digital Sky + UIN + Type Certificate + RPC - the four compliance items for legal commercial drone operation in India.
Green/yellow/red zones - airspace map; green zone allows flight up to 400 ft without permission.
Ultra-low-volume spraying - 10–25 L/acre by drone against 150–250 L/acre by knapsack; droplet density decides efficacy.
Downwash - rotor airflow that drives droplets into the canopy but also increases drift if height or wind is wrong.
Electrostatic and rotary-atomiser nozzles - charged droplets for underleaf deposition; disc rpm controls droplet size.
Section control - automatic shut-off in overlaps and headlands, saving roughly 4–10% of input.
Indian constraint - 1.08 ha average holding, so precision agriculture scales through CHCs, FPOs and drone-as-a-service.

Formulas and examples

NDVI = (R_NIR − R_Red) / (R_NIR + R_Red); dimensionless, −1 to +1 (bare soil ≈ 0.1–0.2, healthy crop ≈ 0.6–0.9)
GSD = (sensor pixel pitch × flying height) / focal length = (sensor width × H) / (focal length × image width in pixels); GSD in cm/pixel, H and focal length in the same unit
Area covered per image = (GSD × image width px) × (GSD × image height px)
Number of images ≈ Area / [image footprint × (1 − front overlap) × (1 − side overlap)]
Field capacity C (ha/h) = (W × S × η) / 10, where W = swath width (m), S = speed (km/h), η = field efficiency (fraction)
Application rate Q (L/ha) = (600 × q × n) / (W × S), q = flow per nozzle (L/min), n = number of nozzles, W = swath (m), S = speed (km/h)
Droplet density N (droplets/cm²) = (6 × V × 10⁷) / (π × VMD³), with V = spray volume in L/ha and VMD in µm (e.g. V = 100 L/ha, VMD = 200 µm → N ≈ 239 droplets/cm²)
Variable rate for a zone: Rate = (Target nutrient uptake − Soil supply) / Fertiliser nutrient fraction × (1/Use efficiency)
Payback period (years) = Capital cost / (Annual saving + value of annual yield gain − annual operating cost)
Spray coverage rate (acre/h) = (Swath × Speed × Field efficiency) / 4047 with swath in m, speed in m/h

Example 1 — GSD and image count for a drone survey. A drone carries a 20 MP camera: sensor width 13.2 mm, image width 5,472 px, focal length 8.8 mm. Flying height 100 m.

GSD = (13.2 mm × 100,000 mm) / (8.8 mm × 5,472 px) = 1,320,000 / 48,153.6 = 27.4 mm/px ≈ 2.74 cm/pixel.
Image footprint width = 5,472 × 0.0274 m = 149.9 m; height (3,648 px) = 3,648 × 0.0274 = 99.9 m.
With 75% front and 65% side overlap, effective footprint = (149.9 × 0.35) × (99.9 × 0.25) = 52.5 × 25.0 = 1,312 m² per image.
For a 20 ha field (200,000 m²): images ≈ 200,000 / 1,312 ≈ 153 images (plus a margin for turns and buffer).

Example 2 — Drone spraying field capacity and cost. Swath 4 m, ground speed 15 km/h, field efficiency 0.70, tank 10 L, application rate 25 L/ha.

Field capacity = (4 × 15 × 0.70)/10 = 4.2 ha/h theoretical-with-efficiency.
Tank covers 10/25 = 0.4 ha per sortie, so about 4.2/0.4 ≈ 10.5 tank refills per hour — in practice battery swaps and refills bring real output to about 1.5–2.5 ha/h, which is why the "1 acre in 7 minutes" figure means flight time only.
At a service charge of ₹500/acre, 1 acre = 0.405 ha. Spraying 10 acres/day (4.05 ha) earns ₹5,000/day; if diesel-generator charging, batteries, travel and operator cost ₹1,800/day, margin ≈ ₹3,200/day.
A ₹6,00,000 drone with 40% subsidy costs the owner ₹3,60,000. At ₹3,200/day margin and 120 working days/year = ₹3,84,000/year → payback ≈ 0.94 year, before maintenance and battery replacement (batteries typically need replacing in 1–2 years, ₹60,000–₹1,00,000).

Example 3 — Variable-rate nitrogen from a zone map. A 12 ha wheat field has three zones. Target 120 kg N/ha for the field; soil tests show available N differs.

Zone	Area (ha)	Soil available N (kg/ha)	N to apply (kg/ha)	Urea @46% N (kg/ha)	Urea needed (kg)
High (good)	5	60	90	195.7	978
Medium	4	40	120	260.9	1,044
Low (poor)	3	25	140	304.3	913

Total urea with VRT = 978 + 1,044 + 913 = 2,935 kg.
Uniform application at 120 kg N/ha = 260.9 kg urea/ha × 12 = 3,131 kg.
Saving = 3,131 − 2,935 = 196 kg urea ≈ 6.3%, worth about ₹1,160 at a subsidised ₹5.9/kg (₹266.5 per 45 kg bag) — small in rupees, but the real gain is correcting under-fertilised low zones and avoiding luxury N where it would cause lodging.

Example 4 — Hydraulic nozzle rate check (calibration). A boom sprayer has 12 nozzles at 0.5 m spacing, each delivering 0.8 L/min at 3 bar; forward speed 4 km/h.

Swath W = 12 × 0.5 = 6 m.
Q = (600 × 0.8 × 12)/(6 × 4) = 5,760/24 = 240 L/ha.
To reach a label rate of 200 L/ha at the same nozzles and pressure, required speed S = (600 × 0.8 × 12)/(6 × 200) = 5,760/1,200 = 4.8 km/h.

Example 5 — Drone vs knapsack vs tractor sprayer for 10 ha paddy.

Parameter	Knapsack (manual)	Tractor boom sprayer	Spray drone (10 L)
Water used	150–250 L/acre	100–150 L/acre	10–25 L/acre
Output	0.3–0.5 ha/day/person	4–6 ha/h	1.5–2.5 ha/h effective
Labour for 10 ha	~20–30 person-days	1 operator, ~2 h	1 pilot + 1 helper, ~5 h
Crop trampling	High	High in tall crops	Nil
Operator exposure	Very high	Moderate	Very low
Capital cost	₹1,500–4,000	₹60,000–1,50,000 + tractor	₹5–10 lakh
Suitability	Small plots, spot spray	Early crop, dry fields	Tall/standing/wet crops, large blocks

Where it is used

Tractor and implement makers: Mahindra (auto-steer, Krish-e advisory), John Deere India (JDLink telematics, guidance), Escorts-Kubota, TAFE, New Holland — roles in product engineering, electronics integration and precision-farming support.
Drone manufacturers and service networks: Garuda Aerospace, IoTechWorld Avigation, Marut Drones, General Aeronautics, Asteria, Thanos — roles as design engineer, agronomy lead, remote pilot, flight-ops and RPTO instructor.
Agri-input and agri-tech companies: IFFCO (nano urea drone spraying), Coromandel, Dhanuka, Bayer, Syngenta, Rallis; CropIn, Fasal, Satyukt, AgNext, DeHaat and Ninjacart for satellite/sensor analytics and advisories.
Government and insurance: ICAR institutes (ICAR-CIAE Bhopal, IARI, CRIDA), State Agricultural Universities and KVKs; Mahalanobis National Crop Forecast Centre and PMFBY crop-cutting/ damage assessment; state horticulture and watershed departments for land levelling and drainage design.
Farm services on the ground: Custom Hiring Centres, FPOs, Namo Drone Didi SHGs and village-level entrepreneurs selling drone spraying and mapping per acre; sugar mills and plantations for cane and tea block monitoring.
Allied use: laser land levelling with GNSS control, precision drip/fertigation design, livestock and pond monitoring, and plantation yield estimation in tea, coffee, coconut and oil palm.

Interview questions

Q: What exactly makes farming "precision"? A: Inputs are varied within the field according to geo-referenced measurement of local need, rather than applied at one uniform rate; the test is whether the decision changes with position.
Q: Difference between map-based and sensor-based variable rate application? A: Map-based needs a prior survey and a prescription map that the controller reads against GNSS position; sensor-based reads the crop or soil in real time and acts in the same pass with no map.
Q: Why is RTK needed when ordinary GPS already works? A: Ordinary GNSS is accurate to a metre or so, which is useless for 2-m rows or auto-steer; RTK's carrier-phase correction gives 2–3 cm, enough for repeatable passes and inter-row operations.
Q: Which NDVI value would you expect from healthy wheat and from bare soil? A: Roughly 0.6–0.9 for a dense healthy canopy and about 0.1–0.2 for bare soil; the index saturates in very dense canopies, where NDRE is preferred.
Q: Why does a drone use only 10–25 litres per acre when a knapsack uses 200? A: Drone spraying is ultra-low-volume with fine droplets and rotor downwash for canopy penetration, so efficacy comes from droplet density and coverage rather than volume of carrier water.
Q: What legal documents must an Indian commercial spray-drone operator hold? A: A DGCA type-certified drone with a UIN registered on Digital Sky, a Remote Pilot Certificate from an approved RPTO, airspace clearance if outside the green zone, plus insurance and compliance with the pesticide drone-spraying SOP and CIB&RC-approved label claims.
Q: Up to what height can you fly in a green zone without permission? A: 400 feet above ground level in a green zone (200 ft within 8–12 km of an airport perimeter as notified); yellow zones need ATC permission and red zones are prohibited.
Q: With holdings averaging about one hectare, is precision agriculture even relevant in India? A: Yes, but delivered as a service — CHC/FPO-owned drones, satellite-based advisories and block-level zoning capture most of the benefit without each farmer owning the equipment.
Q: Why can't a contact fungicide simply be moved from knapsack to drone at the same dose? A: Low-volume fine-droplet application gives different deposition and underleaf coverage, so the label claim, dose per hectare and adjuvant have to be validated — CIB&RC approves drone label claims separately.
Q: What limits a spray drone's practical output per hour? A: Tank volume and battery endurance; a 10 L tank covers about 0.4 ha at 25 L/ha and the battery lasts 10–20 minutes, so refill and battery-swap turnaround, not flight speed, sets the real figure.
Q: Which sensor would you choose to schedule irrigation from a drone, and why? A: A thermal camera — canopy temperature rises when transpiration drops, so the crop water stress index flags stressed zones earlier than a visible or NDVI image.
Q: What is the semivariogram used for? A: It quantifies how soil property similarity decays with distance; its nugget, sill and range set the kriging weights and also tell you the sensible grid-sampling spacing.
Q: How does section control save money without any sensor on the crop? A: It uses the applied-area map and GNSS position to shut off nozzles or rows where the machine has already covered ground or has left the field boundary, cutting overlap waste by several per cent.

Common mistakes

Treating precision agriculture as "using a drone" — the drone is one data-capture and application tool; without a decision rule and differential application, nothing is precise.
Quoting drone coverage from flight time alone and ignoring refill, battery swap and ferry time, which halve the real hectares per hour.
Mixing up accuracy classes: calling a sub-metre DGPS set-up "RTK", or assuming NavIC alone gives centimetre accuracy without a correction service.
Using the uniform-rate dose per hectare for drone spraying instead of the CIB&RC-approved drone label claim and volume.
Forgetting the 600 constant and unit basis in the nozzle discharge formula, or computing swath from the number of nozzles without the spacing.
Reading NDVI as an absolute nutrient value — it is an uncalibrated vigour index, affected by variety, growth stage, soil background and saturation in dense canopies.
Quoting subsidy percentages, approved chemical lists or buffer distances as fixed; these are revised in each year's SMAM guidelines and SOP amendments and must be checked before advising a farmer.
Ignoring wind: spraying above roughly 3 m/s, or at too great a height above the canopy, causes drift that fails the target and endangers neighbouring crops and water bodies.