The rover drives a boustrophedon (back-and-forth) sweep, one crop row at a time — the same lawnmower pattern real autonomous field robots and combine harvesters use to guarantee full coverage with no gaps and no repeated passes.
Weed targeting: d = √((x_rover-x_weed)² + (z_rover-z_weed)²)
fire laser when d < R_scan (R_scan ≈ 0.55 row-widths)
kill after dwell τ ≈ 0.35s / speed× (models laser dwell time)
Field coverage: C = (lanes_done + x_progress / lane_length) / total_lanes × 100%
Drone disease flag: triggers when drone passes within R_air of a
diseased plant while AI scan is enabled (models NDVI / multispectral
imagery classifiers used on real crop-scouting drones).
- Rover speed — scales how fast the rover sweeps each lane; the laser's kill dwell time is divided by this, so a faster rover still needs proportionally less time per weed once it is in range.
- Weed density — how many weeds are seeded across the field; changing it reseeds the field and resets the weed counter.
- Laser weeding — toggles the rover's onboard laser; weeds are only targeted and cleared while it is on, exactly like a Carbon-Robotics-style LaserWeeder.
- Drone AI scan — toggles the overhead drone's disease-detection pass; diseased plants (dark speckled) are only flagged while it is on.
Together the two machines mirror a real precision-agriculture stack: a ground robot handling mechanical or laser weed control lane by lane, and an aerial platform running computer-vision disease detection across the whole field far faster than a human scout could walk it.