The trade fair for Dutch organic farming took place in the Netherlands last week. While there, I talked to one of the initiators of the Odd.bot Weedwacker, a little, autonomous mechanical weeder. In fact, it was more of an invitation to check out its possibilities during a demonstration at Delft Technical University.
Very interesting, but because of a lack of time and mainly because Future Farming already paid attention to it at the end of 2018, I skipped the demonstration. It will be more exciting once the weeder will be demonstrated in the field this spring.
It will be exciting to know whether the robot’s camera can distinguish the different types of weeds from the desired crops. The Odd.bot moves across the field and the camera looks at the different types of plants. Whatever he recognizes as weeds, is destroyed with a sort of chopper.
Organic farmers are very interested in developments like this one, because they cannot spray their crops against weeds and manual labour is not always available. However, the Odd.bot’s developer, in a private initiative together with Delft Technical University, specifically looks at conventional agriculture as well. Chemical weed control is put under more and more pressure there.
Text continues underneath image
In conventional agriculture, the weedbot must enter into competition with autonomous little machines that do not control weeds mechanically. They spray a minuscule amount of herbicide on a detected undesirable plant.
What helps a machine like the odd.bot is that the frequently used pesticide glyphosate is on the verge of a Europe-wide ban. Generally speaking, the desire to avoid compaction of the soil increasingly helps the introduction of light, self-propelling vehicles.
The weedwacker has reached the TRL4 stage rapidly. TRL stands for Technical Readiness Level and is measured on a 1-9 scale. TRL9 means that a system is tested and ready for commercialisation.
As I mentioned before, proof of principle must now become proof of concept. The accurate recognition of weeds, in an early stage as well, is an important aspect. This is difficult for the human eye, let alone for an electronic one. To evade the objection of confusion due to changing light conditions, the idea has risen to mainly work at night and take care of standard lighting themselves. This is a difficult trajectory. “We could sure use some help”, the initiators modestly mention.
Last but not least: what will the costs of a properly functioning machine be? Can it financially compete with manual weeding? Is it feasible to control weeds for about € 250 per hectare?
As said, there are several initiatives in this field and weed recognition is an important aspect. In my opinion, it does not take 10 years before we see the first weed bots scurrying through the fields, just like robotic lawn mowers.