Computer Vision Plant Identification & Navigation System

Autonomous robot using computer vision and SLAM to identify plants in real-time and navigate to them for automated care.

Overview

This project focuses on the perception and autonomy stack for autonomous plant care. The system uses computer vision algorithms to identify plants in real-time, then plans and executes navigation paths. The architecture emphasizes robust perception in dynamic environments and reliable autonomous navigation with SLAM-based localization.

Role

Systems Architect & Lead Developer — Architected the perception pipeline, developed CV algorithms, implemented navigation and path planning, and integrated sensor systems.

Tech Stack

•Computer Vision (OpenCV, TensorFlow)
•SLAM (Simultaneous Localization and Mapping)
•Path Planning (A*, RRT)
•Python
•C++
•ROS2
•RGB-D Camera Systems
•IMU & Odometry Sensors

Key Challenges

•Real-time plant detection in cluttered environments
•Robust visual navigation with limited compute
•Handling occlusions and varying plant appearances
•Integrating perception with motion planning for smooth navigation
•Maintaining localization accuracy during extended operation

Results

•Real-time plant identification with sub-second processing
•Autonomous navigation to identified plant targets
•Robust performance across different lighting conditions and varieties
•Validated through extended autonomous operation sessions

Next Steps

•Deep learning optimization for edge deployment
•Multi-camera fusion for improved perception
•Advanced path planning with dynamic obstacle avoidance
•IoT sensor integration for comprehensive plant health monitoring

Interested in learning more or collaborating on similar projects?

Get In Touch