DRUMLINMOTORS

The fleet

Scout maps, Base plans, and Row Weeder and Alley Weeder confirm each target with their own cameras. One shared record follows a weed through the work.

Watch the fleet work — run Scout and both rovers live, with Base reserving their routes through a full 2.5-acre block.

Map, assign, confirm, record

  1. Scout maps. The quadcopter photographs the block and produces candidate weed locations.
  2. Base plans. Annuals go to Alley Weeder’s oscillating diamond; cockle needs its swappable auger or hand removal; clover, sorrel and smartweed go to Row Weeder’s spot cup; the rest go to hand work.
  3. The rovers confirm. Assignments are places to inspect, never permission to fire. Each rover uses its own camera and crop guards. Sorrel still requires Row Weeder’s operator override; smartweed is suppression, and cockle core recovery needs a hand check.
  4. One ledger follows every weed. Sightings, attempted work, treatment and hand flags stay together whichever machine saw it. Recording a spray does not mean the weed is dead.

The first rover build

Alley Weeder comes first: chassis and cameras for spring image collection, followed by a bench-qualified arm and tool. Row Weeder remains a paper design. The shorter frame-mounted arm serves half a mat from every alley. Two fixed hull cameras find targets while driving; the arm camera confirms before plunging.

The default diamond rocks on a crank with an output slip clutch. Oscillation reduces continuous winding and average twist, but snag loads and actual weed control still need tests. The auger is a manually swappable alternative. Pipe-facing half-mats get a separate pass after people move the pipe; moving and resetting it takes human time.

What the simulations can tell us

Scout supplies places to inspect. It cannot authorize tool use. The live simulations use the current controllers, crop guards and field presets, including charging and pipe-work scheduling. Detection, soil grip, tool effectiveness and timing are estimates, not measured field performance. Earlier whole-mat, stop-and-scan comparisons do not describe this revision.

Read the current tool comparisons and assumptions and the Alley Weeder build notes and parts-list lower bound.

Run Row Weeder’s existing live simulation to see its local controller work through a demo field. Or watch the Alley Weeder work on its own.