Yield & SMS measurement system

[Audio] Welcome at this seminar about digitalization in farming, more specific on potato cultivation and how specialized machinery can be equipped with specific technology to support this. First of all we will do the short introduction on AVR again, to introduce the framework in which AVR as machine manufacturer can play its role..

[Audio] If everything is done properly , last step in the field cycle is to harvest the product. Harvesters can roughly be divided in 2 main groups , more specific trailed harvesters and self propelled..

[Audio] Within trailed potato harvesters we offer different types of machine. Each with its own specific cleaning mechanism, which suit the condition best. Depending on soil type, haulm, variety, desired quality, logistics ….

[Audio] The flagship harvester @ AVR is the Puma4.0. The first puma's revolutionized the concept of 4row self propelled harvesting. This 4th generation was the first connected potato harvester, launched in 2019/20..

[Audio] Now we can get back to our subject : Digitalization implemented in farm machinery. I'll guide you through some possibilities using all steps in some sort of yearly cultivation cycle..

[Audio] Every year the farmer needs to make a cropping scheme. So first it's decided on which plots potatoes will be grown. These fields are analysed and base on analysis a cropping strategy can be made. Which kind of soil preparation, fertilizing, planting distance, kind of seeds , etc will be used. So finally when season starts in spring a suitable soil preparation and fertilizing job is done. After that seeds are planted and the growing can be begin. Typically during the season treatment against pests and extra fertilizing can be done. Some area – more areas every year – provide irrigation. In automn harvesting is done … If we look at where we as avr are in a smart agriculture cycle, we see we can provide agronomical information in 2 crucial stages of the process. First of all : in spring we can provide planting data. And 2nd time is during harvesting, also crucial. Yield measurement is the ultimate evaluation of what happened during season. Did I as a farmer make the right decisions? Do I spot differences within or between fields ? Can I link that variety to choices I made during the season? And offcourse : what do I learn for coming seasons….

[Audio] How can we define smart farming. If we look at the context of farming – definitely arable farming, it's all about working with living material in outside conditions. That's means subject to nature's forces (climate) , subject to pests and diseases , working with specific soil types … And the goal is to maximize a quality yield while optimizing input use and minimizing environmental impact. So it's all about doing the right thing at the right time, and reacting in the best possible way to external influences. Let's see how didgitalization can help achieve this….

[Audio] These are the steps which can be discussed. Let's start..

How can smart – precision – digital agriculture help?.

[Audio] Harvest is the final evaluation – the result of what happened during the season..

[Audio] Harvesting the potatoes at the end of the season is looking at the result of all actions taken. For the tools available we will focus now on self propelled Puma harvester. A potato harvesters takes the ridges with soil, clods, stones and potatoes and a series of cleaning and transport mechanisms bring potatoes to the bunker. If we want to have a geospatial evaluation of the yield, we need to measure on the machine. If we want to product as clean as possible we have to measure on the bunkerfilling band. Depending on conditions the product will be as clean as possible there. In difficult harvesting conditions it's possible that still some soil and clods will be present in the measurend product. We'll talk about that later..

[Audio] The first system we will discuss is the yield measurement system on AVR Puma..

[Audio] The option yield measurement consists of wheighing cells mounted underneath the bunker filling web. You see in the picture where in the machine. It's the last step before the bunker, so before unloading and so best chance to having a clean product so a pure/correct measurement. Because honestly in se it is a stupid system. It just measures everything which passes this web. It doesn't know if it's measuring potatoes – clods – stones – soil – carrots … It just gives you back a weight and time so this can be allocated to a certain geolocation. So this is based on a wheight measured by the weighing cells, considering a speed of the web (measured with speedsensor), a time interval from point of intake in machine to the bunkerfill web (about 23sec) and a speed and working widht of the harvester to show a Ton/ha value….

[Audio] Some important remarks : to link the weight to a correct position in the field it's highly recommended to install a GPS system with RTK correction on the harvester. So a precise measurement can be guaranteed. Just look at the maps mad with and without correction signal and you will understand it will be very difficult to have correct indication of weight if work passes cross so you will have double weights and missing values ….

[Audio] Also important to note is that the system is designed to display yield maps in AVR Connect. What does this mean in practice? You don't have a "real live mapping or indication" while harvesting. The gathered data needs some postprocessing and is then attributed to a field. We can show data and make a map based on a "harvesting trip" during a certain time. But again it is adviceable to have a field prepared in the AVR Connect system, so all data can be automatically attributed to a certain field. Here you see on top fields created by a customer. At the bottom a resulst of a yield map..

[Audio] How does it work? Step 1 is calibrating the system. That's done by pressing "null" – similar to known weighing scales – then the bunkerfilling web will turn for 30s and so a zero weight measurement is done. In this way soil sticking to web or differences in tension of the web can be eliminated. If desired you can add an own estimation in the system of tare you are harvsting. Since this is allmost impossible to do and will probably vary in the field depending conditions, it's really difficult, so adviced is to not do this. But rather do a correction afterwards in the avr connect system for the whole field (if you have official weights of trucks storing the potatoes for example.d Step 2 : harvest !.

[Audio] Extra calibration is possible by weighing tipping trailers if a weighing machine is available at the farm. By weighing trailers and giving correct weight the correction factor will be refined and measurement will be more correct. Don't judge the system on precise kilograms. Seeing and discovering differences in the field also have big value..

[Audio] We do have a live indication but that's just an instant value in ton/ha so it's a value which varies a lot while harvesting. Remember to have a look afterwards in AVR Connect to see a full map and discover insights on what was harvested in the field..

[Audio] Most important is visualisation in AVR connect. Here you see a heat map of ton/ha for each zone in the field. So more or less fertilize zones can nicely be seen. It's possible to change the scale to have more detail in certain interval settings. You can nicely see that at the sides of the field the yield is less then the middle. It's upto the farmer now to see what the reasons can be. Are it wet spots or dry spots? Something went wrong with the planter? Several reasons can be tought off to give an explanation..

[Audio] Export options for further analyses are available. Excel files, CSV files, a shapefile and geojson. When AVR connect is linked to other farm management systems like jd link or dacom , the data can be automatically forwarded to those systems. By exporting data , further analysis by farmer, or agronomist, advicing companies or research groups can be done..

[Audio] The first system we will discuss is the yield measurement system on AVR Puma. This system was introduced to the market end of 2024. It provides interesting information on sizes of harvested product. This being an important quality factor for potatoes. Think french fries , we need as much length as possible. For table potatoes certain sizes are paid better then others. Seed potatoes are also cultivated to obtain certain size classes..

[Audio] This system is based on camera technology. It recognizes potatoes in a product flow apart from clods and or stones. The recognized potatoes are the measured. Length and width are defined. With this data a statistical matrix data set is formed with time stamp so can again be linked to a certain geospatial position..

[Audio] If we look at the hardware we see a "box" is mounted on the bunkerfilling web. This to protect the system from dust and varying light conditions. Within this box e stereocamera is mounted and some led lighting, which create consistent light for correct measurement conditions. Also build in the electrical cabinet is an industrial PC for direct processing of the data..

[Audio] How does the system work? So we mounted a stereovision camera at fixed position above the sieving web. We now the angle in which the camera sees. This camera takes every x seconds a picture. Based on web speed we make sure the pictures don't overlap and we don't have gaps between pictures in that way ensuring a complete measurement is done. All product is seen and no double measurement is done. So now we have the pictures. What's next?.

[Audio] These pictures show a top view of everthing what's underneath. Is sees the same as a huam eye. So the toplayer of product passing the camera; if potato flow is very high and layers exist, only the top layer is measured. Still this gives a very good statistical value of the total crop quality. Then the system needs to recognize potatoes in the picture. This is done with a AI created algorythm. This system still has updates every time we see 'new varieties' or 'new conditions'. The system has 3 classes of recognized potatoes. Green means the system defined the object as a potato, it's free to measure so it is sized (!). Blue : the system defines an object as being a potato but can't measure because overlap or not visible enough to measure. Red objets are defined as potatoes but are not fully in the picture so again a measurement can not be made. So for all green potatoes a bounding box is defined..

[Audio] Here you see an example of potato+clod sample and how the system colours it. you see clods and or haulms are not coloured at all, it nicely shows the system doesn't take into account all material other than potato. And if you look at the blue potatoe in the middle of the picture the system was confused in this situation where 2 tubers where overlapping..

[Audio] So green potatoes get a bounding box. This bounding box is a "best fit" rectangle which can be drawn around the tuber. This rectangle's longest side is the length of the potatoe, shortest side is the width. Width can be seen as the 'size' or 'caliber' of a tuber. A value often referred to in potato business. This data since it all has a timestamp can be allocated to field, but even a very specific location in the field. In this way some representations are possible..

[Audio] How is the data presented?. How is the data presented?.

[Audio] First of all you have heat maps. So geospatial data. What's shown in these maps? A percentage of potatoes with a certain width (can be chosen). For example if for french fries you are paid for potatoes bigger than 40mm , it's very usefull information knowing if certain areas in the field scored higher in % compared to others. Or for seeds if you are aiming for a widht of 35-45mm it's the opposite : here it's important knowing where potatoes grew to big for example..

[Audio] Second heat map shows the length of the potatoes which score above the first boundary. So if you selected 40mm up, now you will have the length of all 40mm up potatoes in certain areas . This can be really intersting information vor selection purposes. If you are looking for as long possible, think long french fries, this gives you this value..

[Audio] Third representation is a Matrix where you see the measurements for a certain field. This will give you a quick 'cloud' in which sizes most harvest crop can be allocated. In this example you see big differences. A nice percentage is over 60mm in width. But also less then 45mm is over 30%. So geospatial analysis will be necessary to see if this can be appointed to certain conditions in the field or if it is a heterogenous harvest all over the field..

[Audio] Some final remarks on this system : It gives a nice statistical overview of harvested potato sized ! Width and length ! Only potatoes are measured so data is only on potatoes no measurement errors are made there, as opposed to weighing system, where it's possible that badly harvested parts of the field measure high yield, since soil is more heavy then potatoe. BUT beware ! There is also no information on amount of clods or soil, that's off course the dream to have a precise estimation on tare so a corrected yield measurement could be done like that. System sees a top view , so at high yields, full sieving webs the camera doesn't measure every tuber, so again it's a statistical indication for what's harvested in the field..

(b). Figuur 9: roepassjng van de R6B meth0de op toto fiddLA09 toto R6B-camera en (b) R6B geannoteerde roto..

[Audio] That's about all I got for now. I hope you got some interesting insights how technology on machinery combined with digital information can bring agricultural practices forward. And please stay tuned via socials to follow what we launch next. For any questions or proposals for collaborations, please contact via e-mail..

Yield monitoring. Harvesters provide detailed information on actual yields extensive coverage, throughout the EU Main objective of this Lab: to unlock the potential of yield data from harvesters for European-wide yield monitoring 2 key issues will be addressed: access to often very scattered harvester data, while respecting the privacy of the data owners (farmers) using the harvester data for yield monitoring to estimate crop productivity at local scale and at regional scale throughout the EU, considering the different growing conditions.

Thank you for your attention!.

STAY TUNED!. A black and white circle with icons Description automatically generated A black and white circle with icons Description automatically generated A black and white circle with icons Description automatically generated A black and white circle with icons Description automatically generated A black and white circle with icons Description automatically generated.