Machine Vision
Orchard scenes contain overlapping plants, shadows, and changing viewpoints. I developed a vision pipeline combining weed segmentation, identity tracking, and spatial analysis to locate individual weeds and estimate intra-row density.
Research
Computer vision, robotic control, and plant-informed decisions for precision agriculture. Explore the projects and watch the systems in action.
Mechatronics & Control
A detected weed must become a physical spray target. I designed and evaluated a three-nozzle servo-controlled system that translates target locations into nozzle assignments, steering commands, and spray timing, with laboratory validation of actuator and targeting performance.
Decision Intelligence
Weeds differ in canopy size and growth habit. I developed a calibrated imaging workflow to measure plant morphology and used controlled spray-deposition experiments to examine how plant and nozzle characteristics affect coverage and off-target deposition.
Systems Integration
Perception and spraying must work together while the robot moves. I integrated target localization, spray decisions, and nozzle operation on a Farm-ng Amiga platform, with orchard field tests and an interface for live monitoring and control.