The 2022 growing season saw the arrival of a new crop camera on arable farms in the Netherlands, the
Augmenta Mantis. This multi-spectral camera is mounted on the roof of the tractor. The camera scans the crop or soil 20 metres forward and 40 metres wide in high resolution.
The crop camera costs around €17,000, and creates biomass images in no time. The device can present the full-width image in 128 separate tracks. The camera can be used for real-time precision applications, where the leaf mass of crop or weeds is normative, as well as for scans that can be used to track crop development or worked into prescription maps.
Full multispectral camera image
The Augmenta Mantis can be seen as a substantially improved greenseeker, like the Yara N-sensor or Fritsmeyer. All look ahead from the tractor, scan the crop and produce an image of the density of the vegetation. A big difference is that the Mantis gives a full multispectral camera image over the whole 40-45 metres. Currently, several dozen systems have now been sold in the Netherlands.
Experiences of NPPL participants
Several participants in the ‘National Precision Farming Laboratory’ started working with the Augmenta Mantis system last summer. The crop sensor is seen as an asset in (data collection for) precision agriculture. In terms of working under difficult light conditions, reliability of the image of leaf mass and data processing capabilities, the Dutch growers say it still needs improvement.
Brian Salomé: ‘Augmenta should also be able to make prescription maps’ Brian Salomé. - Photo: Peter Roek Brian Salomé has had the Augmenta Mantis crop camera on the roof of his John Deere tractor with mounted sprayer since September of 2022. When spraying his potatoes, he has used the camera and associated software to variably spray his potato foliage. That is; making the dosage of Quickdown depend on leaf mass. Salomé took the recommended dosage of 0.8 litres per hectare as a base and varied it by plus or minus 10%. “Starting with 250 litres of water, I varied between 220 and 280 litres of water per hectare. We worked in real-time; the camera then scans the crop in front of the tractor and based on that, the tablet in the tractor controls the sprayer.” Working in real-time was a necessity; currently it’s not possible to create prescription maps via MyJohnDeere based on the crop scan with the Augmenta sensor. “That needs to be fixed first next winter. That’s why I have the sensor on trial for now. MyJohnDeere cannot read the information from the sensor at the moment. I will also get support from NPPL on that.” Salomé feels that the system should also be able to provide him with prescription maps to enable variable-rate spraying at any time of the day. “Right now, when it’s a little too dark outside, images can’t be created. That is difficult, but could be solved by making a scan of the field in the middle of the day beforehand, working it out in a prescription map and spraying based on that in the evening or the next morning.” Another disadvantage of working in real-time is you don’t know in advance exactly how much herbicide you will need. “That is quite tricky. You either have some left in your sprayer or you come up just a little short at the end of your plot. I solved this for haulmicide by filling the tank for half an acre less for the 5-hectare plot. That worked out reasonably well, but it’s not how precise you actually want to work.” For next year, Salomé plans to use the Augmenta sensor for variable rate herbicide dosing in different crops, as well as in nitrogen fertilisation of his wheat. Variable fertilisation of potatoes is another story. Salomé combines this with spraying against phytophthora. “Then you run into a problem: in a spot with a lot of added leaf mass, you would want to fertilise a bit less, but on the other hand in such a spot you would want to apply more phytopthora agent. To solve this, you have to think of injecting foliar fertiliser into the spray line based on a presciption map, and then applying the pesticide in real-time or using a second prescription map in such a way that you can vary spot by spot.” Salomé estimates he can recoup the investment of some €18,000 in the sensor in roughly five years. “That’s €3,600 a year. This should be possible if I can also use it to make prescription maps. Having those made for me costs €20 per hectare.” Pieter van Leeuwen Boomkamp: ‘Confidence in the technology, a few imperfections though’ Pieter van Leeuwen Boomkamp. - Photo: Koos Groenewold Before Pieter van Leeuwen Boomkamp decides to buy the Augmenta Mantis crop sensor, at least two things need to be improved. 1: The medium dosage based on the determined leaf mass should not be able to be reduced by 50%, but by 100%. 2: The sensor must become more light-sensitive than it is now. “In mid-October on a cloudy day, it was actually too dark before ten in the morning and after four in the afternoon for a reliable crop scan.” “I sprayed off potato haulm quite late, in October. By then, there are already many bits in the plot where there are no leaves left. I don’t want to spray anything there then, that’s when the biggest savings are made. With BBleap’s system (with PWM technology, ed.) that I have on the sprayer, I should be able to vary between 100 and 0%. So far, however, this has not been possible.” Van Leeuwen Boomkamp harvested his starch potatoes towards the end of October and sprayed them with Quickdown in mid-October. In doing so, he set the maximum dosage at 0.6 litres per hectare (the standard recommended dosage is 0.8 litres). From there, the dosage then went down to a minimum of 0.3 litres per hectare. “That minimum should have been zero,” he says. The second thing Pieter van Leeuwen Boomkamp would like to see improved is Augment Mantis’ light sensitivity. “In autumn, the sensor appears to be usable for only a limited part of the day. That’s a limitation anyway, but especially for my situation. I prefer to spray in the late evening or at night. Then there is less evaporation and ascent of the spray mist by warm air and generally less wind. Moreover, working in the very early morning is nice because you’re not disturbed by the phone. Anyway, that moderate light sensitivity of the system is a limitation. That has to change. I have heard that an update is on the way that would reduce this problem.” Apart from the imperfections mentioned, Van Leeuwen-Boomkamp says he does have confidence in the sensor technology and the system based on his experiences. Apart from the data he needs for variable rate spraying or fertilising – which he plans to do in 2023 – it allows him to collect large amounts of data from his crops. As for the ROI – Van Leeuwen Boomkamp is still in doubt about the purchase – he expects not only to save money on pesticides, but also a reduction in crop damage because less spraying is needed. A payback period of five years, mentioned by his colleague Salomé, seems a bit on the short side in his situation, “but actually a prerequisite”. Pim Sturm: ‘Some teething problems, but sometimes exceptionally good’ Pim Sturm. - Photo: Lex Salverda For contractor Pim Sturm the Augmenta Mantis crop sensor is a step towards real-time variable rate spraying and fertiliser spreading. Other precision applications include variable application of growth regulators and fungicides. ”We are working on collecting data that will allow us to control machines in real time. We then skip the step of loading images and creating prescription maps ourselves.” In terms of detail, the images still lag behind shots taken from a drone, Sturm says. And the effect of the position of the sun is also present here. When the sun is lower, scanning sometimes stops. This is inconvenient as Sturm likes to take advantage of favourable spraying conditions in early morning and late evening. The alternation of the sun and clouds also can be a problem. “One moment the sun is shining, a moment later a cloud passes by. That leads to very strange data. No analysis can be done based on that. As a result, only about 10% of the images we took with the sensor ended up being usable. 80-90% was unusable.” Another imperfection Sturm has encountered is the disruption of the imagery in onions caused by wide wheel tracks. “Somehow the computer averages the image in the tracks, where there is no leaf mass, with the part in between, where the onions do stand well. The result is that for such a 2.25-metre bed, the machine judges that there is very little leaf mass and therefore wrongly increases or decreases the dose there. To solve that, you would have to exclude wheel tracks beforehand.” Pim Sturm, on the other hand, is also keen to say that in a full crop and in constant weather, the Augmenta sensor provides “an extremely good picture” of what is in the field and the variation in it. “Which, by the way, can then often be traced back to the variation in the soil, so to the soil maps.” Sjaak Huetink: ‘Much better than the greenseeker, but far from perfect’ Sjaak Huetink. - Photo: Koos Groenewold Arable/bulb grower Sjaak Huetink finds it difficult to pass judgement on the Augmenta Mantis crop sensor at this stage. “We are in a learning process. The Augmenta is much better than the greenseeker, indicates variation better, but it is still far from perfect.” Among other things, Huetink has used the camera in spot variable dosing of fungicide in onions. This goes well as long as the crop is green. However, when ripening starts and the first yellow spots appear, things start to go wrong, he says. “As soon as the image is no longer uniformly green, you no longer get a reliable picture of how the crop is doing.” Scanning lilies also proves tricky in that plant sizes vary quite a bit. “A fine-sized crop leads to a completely different picture than a coarser crop.” Another issue is the effect of the rising and setting sun. When there is enough sunlight, good images are created. When there is too little sunlight, no images are created. In between is another phase where scanning is done, but the images are of too low quality. “That fits poorly with our desire to spray under optimal conditions,” he says. Sjaak Huetink would also like to see some improvements when it comes to processing of data. Ideally, the weekly – huge – amounts of data he collects should be stored away overnight in his own management programme. “That would make the data more accessible than just through the Augmenta portal and would allow me to use my own calculation methods,” he says. Bram Veldhuisen, Wageningen University & Research: ‘Gain experience first, you don’t trust a new crop advisor right away either’ Bram Veldhuisen. - Photo: Stadje Media At the Wageningen University & Research (WUR) Farm of the Future the tractor for the mounted sprayer was equipped with the Augmenta Mantis sensor last autumn. Apart from an occasional tour of the field, the sensor has not yet been used. The Hardi sprayer was also equipped with the BBLeap PWM system that allows the nozzle output to be varied between 0 and 100% at a constant pressure. Bram Veldhuisen, researcher precision agriculture and robotics at WUR, says 2023 is first and foremost going to be a learning year. “We will first collect data and look at what data the sensor collects, and how that matches what we see in the field. We will also test the reliability of the system. Compare it to a new crop consultant on your farm – you first look at what he has to offer, you don’t blindly follow his advice from day one.” Veldhuisen says the sensor will not only be used for spraying. “In fact, we do as little spraying as possible here. But even with hoeing or any operation with this tractor, we make a crop scan and collect data, which we may be able to do something with sooner or later.” Apart from variable spraying, the Augmenta sensor can also be used for site-specific variable dosing of fertiliser, slurry and solid manure. “The sensor can control any Isobus machine,” says Veldhuisen. Supplier Vantage Agrometius responds Harold Zondag. - Photo: Anne van der Woude Harold Zondag of Dutch supplier Vantage Agrometius responds to inadequacies named by users of the Augmenta Mantis. The Augmenta Mantis, due to lack of light, does not work at precisely those times when many arable farmers like to spray, early in the morning or in the evening.
“True,” says Zondag, “and we have never claimed otherwise. There is a difference with sensors of for instance the Yara-N type with an active light sensor. With that, you can also work in the dark. Augmenta is also working on that. The first experiments indicate that there are good possibilities to achieve results with light of a limited wavelength, the red edge spectrum. In doing so, more differentiation should then be achieved in full green crops, as saturation occurs less quickly. This should be ready by the end of 2023. It is not expected that this light will be as wide as the 40 to 45 metres width that the sensor can handle. It is still a limitation now, though, so it is something that will have to be done at some point. It’s coming soon.”
With My John Deere, prescription maps cannot be created based on the Augmenta Mantis data.
“We checked that in the United States, where the camera is widely used, and indeed you cannot get the shape files read into My John Deere. However, there are several solutions to this limitation. With a small diversion via another programme, like Trimble’s Farmworks, it can be done. Data from the Augmenta sensor can also be imported via WUR’s Farmmaps platform.”