All projects
03
Embedded & IoTCompleted
Embedded · Cloud · AI
Automated Indoor Agricultural System
A full-stack embedded system connecting ESP32 microcontrollers and cameras to a cloud backend. Sensors monitor the growing environment, actuators control equipment, and a computer-vision model evaluates plant health to drive automated responses.
Status
Completed
Timeline
2025
Role
Engineer
Category
Embedded & IoT
Technology Stack
ESP32ESP32-CAMAWSFastAPIPythonResNet18Embedded SystemsSensorsComputer Vision
Problem
Indoor growing environments need constant monitoring and precise control that is impractical to do manually.
Constraints
- Low-cost embedded hardware
- Intermittent connectivity
- Real-time actuation
Solution
Networked ESP32 nodes stream sensor data and images to a cloud API that runs a plant-health model and returns control decisions.
Key Decisions
- Offload inference to the cloud to keep nodes cheap
- Stateless API for easy scaling
System Architecture
01ESP32 sensor/actuator nodes
02ESP32-CAM imaging
03Cloud ingestion API (FastAPI)
04AI inference (ResNet18)
05Automated control loop
Diagram placeholder — replace with a detailed architecture diagram when available.
Hardware
- ESP32
- ESP32-CAM
- Environmental sensors
- Relays / actuators
Software
- FastAPI
- PyTorch (ResNet18)
- AWS
- Python firmware tooling
Challenges
- Reliable image capture on constrained hardware
- Robust connectivity handling
Results
- Working prototype demonstrating closed-loop monitoring, control and AI assessment.
Lessons Learned
- Clear boundaries between edge and cloud simplify embedded systems.
Media
Media placeholder
Media placeholder