The future of non-chemical weeding technology...

Non-chemical weeding is one of the hot topics of ag-tech with many start-ups clambering for investment in their concepts and ideas. Some of this drive has come from market push back on the use of agrochemicals, with their use coming under increasing fire from lawsuits in the US, environmental groups and changing consumer trends.

Some of the concepts of non chemical weed removal are nearly as old as agriculture itself, humans have been pulling up weeds and controlling them with mechanical methods for thousands of years.

While other methods may seem altogether more futuristic, even lasers were trialled in research projects as far back as 1970’s and electric ‘zappers’ have been conceived of since the 1800’s.

To better understand non-chemical weeding it is good to first understand some of the challenges:

Lack of data - Designing any new product or system requires a very good understanding of the problem. Gathering data on weed population and distribution has historically been VERY labour intensive, leaving such high resolution geospatial mapping work as the preserve of large scale research projects.

Scale - Agricultural production is necessarily large scale, even a low weed density of a handful of plants per metre squared becomes a scarily large overall weed population of hundreds of thousands or millions of plants to be dealt with on a field or farm scale. For example the 'John Nix Pocket book for farm management' by The Andersons Centre states that a population of 12-25 Blackgrass plants per square metre could cause a yield loss of 5-15% in a wheat crop. Extrapolating 25 plants per square meter across 20 ha field would mean up to 5 million Blackgrass plants to be controlled! This is an overly simplified model but it does help to put the scale into context.

Energy - Carrying out any action to control a weed has an energy cost, a cost that will be multiplied across the area to be treated. This can lead to the need for serious amounts of energy storage, think very large LPG gas bottles for flame weeding or steam systems, huge batteries or PTO driven electric generators for ‘zappers’ like those produced by Zasso Group and RootWave

Speed - Some non-chemical methods are very time consuming to employ in the field with painfully slow coverage rates when compared to a typical blanket herbicide application. Even with a weed targeting system the speed can be very limited. A treatment frequency of 1hz (1 weed treated per second), perceived by an observer as fast, can still be too slow. For context a population of one hundred thousand weed plants would take over 24hrs to treat at a treatment frequency of 1hz.

Residual control - Most non-chemical weeding technology offers no protection from reinfestation. Leaving the field or crop unprotected while the weeding machine is not operating, leaving later emerging weeds untouched. Whereas some chemicals are able to continue to control weeds for some time after the sprayer has completed its passes.

Collateral damage - Some methods can cause disturbance to the crop, mechanical methods can cause root damage to sensitive crops and other methods like flamers are indiscriminate anything that passes under a flamer will be killed.

“One trick pony” - Dedicated non-chemical weeding machinery can be limited to only weeding, potentially making it a hard sell to farmers when compared to a sprayer that can deliver a diverse ‘toolbox’ of capabilities to a farmer, for the control of pests, fungal problems and growth rates.

Many of these challenges have been around for many years, so what is changing?

Machine learning and AI - The combination of vision systems and Edge AI that can be trained to decide what is and isn’t a weed has been a game changer for precision chemical application sprayers such as the John Deere 'See and spray' system. As the saying goes 'a rising tide lifts all boats' and so this same approach has been making selective non-chemical based treatments possible too, with examples such as Carbon Robotics and Ekobot - the evolution of agriculture showing what the state of the art looks like for targeted non-chemical weed control.

Robotics - The ability to move a treatment method (be it laser, electrical zapper or mechanical hoe) with millimetre precision to a selected target has only been possible with advancements in robotics, fuelled by cheap and easy to program hardware such as Arduino boards. Alongside the development of multipurpose actuators from companies like LINAK and the reduction in the cost of brushless motor and stepper motor technology.

Automation - Removing humans from tedious tasks like weeding either by hand or driving a tractor helps to tackle the residual control problem, as systems can make repeat visits to the field to remove weeds for a significantly lower cost. Retrofit companies such as Bluewhite and Sabanto are showing how automation can be applied to existing machinery and AgXeed and AGROINTELLI are showing what dedicated autonomous machinery can achieve.

Battery technology - For electrical systems batteries are becoming more viable as an option weed control tasks. This has in part been pushed by the advancements in electric vehicles as well as personal devices like laptops and phones. Companies such asVARTA AG are showing that the agricultural space is worthy of investment and have been actively pursuing agricultural applications for their lithium battery technology.

GPS - Being able to get centimetre level accuracy for farm machinery isn’t new any more, farmers have been using GPS on tractors for years. However the relatively recent merger between AGCO Corporation and Trimble Inc. shows that GPS is a very valuable asset to the traditional tractor manufacturers. Combing GPS capability with AI and robotics is a potent mix and a key part of many weeding system technology stacks.

Crop scouting and mapping with drones such as the Drone Ag Limited Skippy Scout system is becoming more mainstream providing a better understanding of weed distribution and population on farms. This kind of data is vital to those companies trying to develop novel weeding methods. It helps to define the scale and distribution of weed populations that any new system will have to be able to cope with.

Funding - Recently the UK government announced a funding package (Farming Equipment and Technology Fund 2024) aimed at helping farmers to improve their productivity. This fund provides money to purchase existing technologies such as inter row weeding, camera guided spray systems and combine mounted hammer mills for weed seed destruction during harvest (among many other pieces of machinery). By providing funding to farmers the UK government will boost sales of such machinery and help existing manufacturers to develop their product further.

Niches - Non-chemical weeding technology is taking a real foothold in some of the niche areas of agriculture where the crops are high value and the only alternatives are expensive hand weeding. Niches like this are helping to build capabilities and drive down technology costs, opening up opportunities to adapt existing tech to suit broadacre crop applications. A case-in-point is GARFORD FARM MACHINERY LIMITED that has been manufacturing machine vision driven solutions for mechanical weeding for horticultural niches for over 20 years and are now researching the deployment of their tech into cereal crop production.

So, what will the future of non-chemical weeding look like?

The opportunities in non-chemical weeding technology are exciting to farmers and technologists alike. The future of the space is likely to be very diverse with specific tech for different crops, regions and farming methods. However there are some key technologies such as Edge AI and machine vision that will likely provide the backbone to most systems.

With the steady march of Moore's law increasing processing power and computing speeds, a successful system will be able to act as quickly as it can ‘think’ in order to achieve a commercially viable treatment frequency. This speed will need to be coupled with a total system energy draw and energy storage capability that can work in remote locations for extended periods of time.

The final piece of the puzzle will be how the machinery will be deployed on farms. Carbon Robotics famously pivoted from a dedicated weeding robot to a tractor mounted tool it is likely that future systems will develop from tractor mounted tools, making adoption and integration for farmers as easy as possible. Dedicated weeding machinery or new concepts like swarm robotics may be adopted as evolutions of the established tractor mounted machinery.