4 Capstone Projects: Agriculture and Environment

4.5.1.5 System Architecture

Data and command flow:

Step 1 — Field node samples soil moisture, soil temperature, water level, and ambient conditions. Step 2 — Node publishes readings over LoRa or Wi-Fi. Step 3 — Gateway or cloud service stores readings and evaluates watering rules. Step 4 — Safe watering commands return to the node, where local firmware checks final safety constraints before actuating the pump.

4.5.1.6 Smart Watering Algorithm

def should_water(soil_moisture, weather_forecast, last_watered):
    """
    Intelligent watering decision algorithm.

    Args:
        soil_moisture: Current soil moisture (0-100%)
        weather_forecast: Dict with rain_probability, temp
        last_watered: Datetime of last watering

    Returns:
        (should_water: bool, duration_seconds: int)
    """

    # Define thresholds (plant-specific)
    DRY_THRESHOLD = 30      # Water if below this
    WET_THRESHOLD = 70      # Never water above this
    RAIN_THRESHOLD = 60     # Don't water if rain likely
    MIN_INTERVAL_HOURS = 6  # Minimum time between watering

    # Check if too soon since last watering
    hours_since = (datetime.now() - last_watered).total_seconds() / 3600
    if hours_since < MIN_INTERVAL_HOURS:
        return (False, 0)

    # Check if rain is expected
    if weather_forecast['rain_probability'] > RAIN_THRESHOLD:
        return (False, 0)  # Let nature do the work

    # Check soil moisture
    if soil_moisture > WET_THRESHOLD:
        return (False, 0)  # Already wet enough

    if soil_moisture < DRY_THRESHOLD:
        # Calculate watering duration based on how dry
        deficit = DRY_THRESHOLD - soil_moisture
        duration = min(deficit * 2, 60)  # Max 60 seconds
        return (True, duration)

    return (False, 0)

4.5.1.7 Implementation Milestones

4.5.1.8 Evaluation Rubric

Read the rubric as a single field-deployment argument. Calibrated sensors establish the evidence, watering control acts on it, power and connectivity keep the loop available outdoors, the algorithm explains when action is permitted, and documentation preserves calibration, safety, and recovery steps. No category can compensate for an unsafe or unreviewable control path.

4.6.2.5 Architecture Design

Data flow:

Step 1 — Sensors provide temperature, humidity, CO2, particulate, light, and noise readings. Step 2 — ESP32 validates and publishes readings over Wi-Fi using MQTT. Step 3 — Cloud services store time-series data and evaluate alert thresholds. Step 4 — Dashboard presents current status, trends, and alerts for facility staff.

4.6.2.6 Implementation Milestones

// Milestone 1: Basic sensor reading
void loop() {
    float temp = readTemperature();
    float humidity = readHumidity();
    int co2 = readCO2();
    float pm25 = readPM25();

    Serial.printf("T:%.1f H:%.1f CO2:%d PM2.5:%.1f\n",
                  temp, humidity, co2, pm25);
    displayOnOLED(temp, humidity, co2, pm25);
    delay(5000);
}

environment/
  +-- {device_id}/
  |   +-- temperature
  |   +-- humidity
  |   +-- co2
  |   +-- pm25
  |   +-- light
  |   +-- status
  +-- alerts/
      +-- {device_id}

4.6.2.7 Evaluation Rubric

4.6.2.8 Extension Ideas

Treat each extension as a new contract, not a decorative extra. Voice alerts add accessibility and notification-state questions; occupancy learning adds training and validation evidence; a mobile client adds authentication and stale-state handling; multiple rooms add routing and mesh reliability; and building-management integration adds an authority boundary that must fail safely.

Add voice alerts via speaker Implement machine learning for occupancy detection Create mobile app for monitoring Add multiple rooms with mesh networking Integrate with building management systems