Outside aerospace · Design Thinking Lab · 2024–25

Cloud-based irrigation to stop over-watering

Team Hypernova · AS247DL / ME247DL · submitted 10 June 2025
With Pranathi S, K M Talin Thimmaiah, and Len Achaiah K M. Faculty: Dr. P. R. Venkatesh, Dr. Ramakrishna Hegde, Dr. Ramaa, Prof. Mukesh M.

Most of India’s freshwater goes to farms, and a lot of that is wasted on a timer. This is a small loop that waters only when the soil is dry and rain is not coming.

85.7%still water by hand
ESP32on Wi-Fi
24 hrain forecast
manualpump override

Capacitive moisture in the ground, OpenWeatherMap for a 24-hour rain call, ESP32 on Wi-Fi, a relayed pump, and a Blynk dashboard with a switch you can actually hit. I worked procurement and electrical.

The enclosure is a SolidWorks part, 3D-printed in red, with the probe sticking out of the bottom so it can be pushed into a pot or a bed.

Prototype demo. Phone portrait, as shot.

The problem we actually heard

A Google Form plus observation, ~70 responses. 85.7% still water by hand. 75.4% wanted semi-automatic, not a fully closed loop they cannot touch. Cost (62.9%) and software reliability (54.3%) beat “no concerns.” That is why the dashboard has a pump switch, not only an algorithm.

Survey: 85.7 percent still water manually — How people irrigate today. Manual is still the default.

Survey concerns: cost, reliability, internet — Concerns about cloud irrigation: cost first, then software, then connectivity.

What we built

Controller	ESP32 / NodeMCU, Arduino IDE, Wi-Fi to the cloud.
Sense	Capacitive soil-moisture probe (FC-28 style in the circuit).
Actuate	Relay module + 5 V / 12 V pump.
Weather	OpenWeatherMap. Skip watering if rain is expected in 24 h.
Failsafe	A second, lower moisture threshold turns the pump on anyway if the forecast was wrong and the soil keeps drying.
UI	Blynk — “Hypernova Agri Solutions.” Moisture, rain probability, pump ON/OFF, temperature, Bengaluru as the city.
Decision concept	The write-up also specifies a gradient-boosting model on soil type, crop, temperature, and weather. The hardware we shipped is the threshold + forecast + failsafe loop.

3D-printed red probe housing with soil electrodes — Printed housing and probe. Push it in, power it, read moisture.

SolidWorks model of the probe housing — SolidWorks enclosure: box, probe boss, room for the relay and ESP.

Circuit: NodeMCU, relay, soil probe, pump, Blynk and weather API — Wiring: MCU, relay, FC-28 probe, pump. Blynk and the weather API sit on the Wi-Fi side.

Blynk dashboard showing moisture 100 percent, rain yes, pump off — Blynk at 22:57, Bengaluru. Moisture 100%, rain yes, pump off — which is the point of the rain gate.

Prototype in a soil cup with relay LED on — Bench prototype in a soil cup. Relay board on the side of the print.

How it decides

Moisture below the crop threshold and no rain in the next day → pump on. Rain forecasted → wait. Moisture still falling through a lower failsafe → pump on regardless. Cloud log: moisture, weather, irrigation events. Manual override on the dashboard because 75% of the survey asked for it.

The report’s objective was at least 30% less water than a timer. Testing with small-scale users and gardeners confirmed savings, less labour, and fewer pre-rain cycles. We did not publish a single farm-scale percentage, so I am not inventing one here.

What we would do next

Mobile-first UI and SMS for phones that are not smartphones.
Offline / edge decisions for rural links (the survey’s connectivity worry).
NPK and pH if the same stick is going to talk about fertiliser, not only water.
LoRaWAN or GSM instead of only Wi-Fi. Solar for off-grid beds.

← Projects