Reference · Agricultural robotics
Electronics for a machine that spends its life outdoors
Paltech builds autonomous robots that take weeds out of grassland without a drop of herbicide. Attacan develops the electronics underneath: the power distribution, the wiring, and the path from a working design to something a factory can build repeatedly.
Farming without the chemistry
Paltech started on an Allgäu family farm that had converted to organic, where dock in the grassland is a genuine problem and the usual answer is a herbicide nobody organic is allowed to use. Their grassland robot, PRATUM, finds the weed, removes it mechanically, and reseeds the gap so the turf closes again.
That is a machine doing patiently and precisely what a person with a spade would otherwise do across a whole meadow. It only works if the robot runs long enough, often enough and reliably enough to cover real ground. Which makes it an electronics problem as much as an agricultural one.
Paltech is in Kempten and Attacan in Pfronten, half an hour apart. That matters more than it sounds. When something misbehaves in a field, somebody can be standing next to it the same day instead of describing it down a phone line.
Years outdoors, not an afternoon
Every robot runs on batteries, so that is not what makes this hard. What makes it hard is where the machine lives. It works through summer heat and winter cold, through rain and snow, and it sits outside between jobs rather than in a heated workshop. Condensation forms inside enclosures when the temperature swings. Seals that hold in June get tested again in February.
And a farmer does not buy a machine for one season. A mower or a loader is expected to still be working in a decade, and an autonomous robot lands in the same expectation whether or not its designers planned for it. That changes what "good enough" means at every level: which connector you specify, how a cable is routed so it does not chafe over thousands of hours, whether a part can be replaced in a yard with ordinary tools.
So the electronics are designed against a service life rather than a demo. Components get judged on how many times they will be mated over years, what vibration does to them across uneven ground, and what happens when the machine is washed down at pressure. That is the difference between hardware that survives a field trial and hardware that survives farming.
Power that has to share a frame with everything else
The robot carries drive motors, sensors, navigation and radio in one chassis, all fed from batteries with no mains within reach. The motors draw hard and switch fast. The sensing and comms electronics need a quiet supply. They sit centimetres apart.
Attacan split the job across layers instead of one board doing everything: a main power unit, a logic board that decides what gets energy and when, and breakouts that carry it to the far corners of the machine. The high-power and low-power harnesses run as separate looms, so switching noise from the drives does not end up in the sensor lines.
The first design was wrong in places, and the field said so
It usually is. A robot that behaves on a bench meets dust, damp, vibration and temperature swings the moment it goes outside, and some of what looked settled stops being settled.
Rather than patch around it, the power distribution went through a second generation. Failures were traced to causes instead of symptoms, hardware and firmware were brought back onto one version, and the documentation was rewritten so the next person to open the machine can follow it.
This is the part worth reading, because it is what the collaboration is actually for. Anyone can deliver a design. The question is who is still there when the field disagrees with it.
Designed so it can be built more than once
A prototype that only its author can assemble is not finished. Attacan structures the bill of materials for real sourcing, specifies the cable sets so they can be made repeatably instead of hand-fitted, and prices and sources the mechanical substructure through its own manufacturing network. Prototypes get built and documented for handover, which is where most of the friction in a hardware project quietly lives.
What Attacan does here
Power distribution
Layered architecture, from the main unit through the logic board to the breakouts.
Design for service life
Components and wiring specified against years of weather, vibration and washdown rather than a trial season.
Redesign
Field failures traced to causes, with hardware and firmware brought back into step.
Manufacturing
BOM structure, sourcing, mechanical substructure, prototype builds and handover documentation.
Hardware that has to survive where it works?
We discuss every engineering challenge in detail before proposing anything.
Discuss your engineering challenge →