AI Smart Plant Watering System based on ESP32-S2

Client

An indoor gardening startup creating a precision irrigation system for home herb gardens and potted plants.

Challenge

Building a system that can distinguish between 'soil needs water' and 'temperature caused sensor drift' without manual calibration per plant species.

Product Concept & Architecture

The ESP32-S2 drives a capacitive moisture sensor array and a DS18B20 soil temperature probe. A 12V latching solenoid valve controls drip irrigation. The ESP32-S2's native USB supports easy field firmware updates.

Engineering Solution

We developed a self-calibrating algorithm in ESP-IDF that tracks baseline dry/wet resistance curves per plant. A lightweight MLP (Multi-Layer Perceptron) classifier determines optimal watering timing based on moisture trend, temperature, and the fetched 3-day weather forecast. The system uses MQTT over Wi-Fi to sync with a companion app. A watchdog timer and brownout detector ensure reliable long-term unattended operation.

Result

98% accurate soil moisture readings across 15 plant species. 40% reduction in water usage vs. timer-based irrigation. Zero false watering events during rain forecasts.