fanquake

"nor do they recognise the importance of having robust netcode"

Refactoring of the core networking code has been ongoing for quite some time. See https://github.com/bitcoin/bitcoin/projects/4.

"Can you name anybody with a good background in networking? You can't point to a list of 100 odd names and assert that certain skills must be present"

sipa? He worked as an SRE at Google before starting work on bitcoin. [1]

rustyrussell? He has done extensive work on the linux kernels networking subsystems. [2]

cdecker? He did a phd at the Distributed Computing Group in Zurich which included work on scaling bitcoin. [3]

gwillen? Also a SRE at Google before bitcoin, and worked on Dapper[1].

[1] https://www.linkedin.com/in/pieterwuille

[2] https://en.wikipedia.org/wiki/Rusty_Russell

[3] https://disco.ethz.ch/alumni/cdecker

[4] https://static.googleusercontent.com/media/research.google.c...

It's not observing itself, it's being observed by other autonomous vehicles.

There are sensors designed for broadacre cropping that will pick weeds out of crop.[0]

As well as sensors that will pickup weeds from dirt.[1]

This would be in the middle of those two problems. There isn't a dense crop to pick the weeds out of, but it's probably slightly too dense to just pick them up from the dirt.

[0]http://www.spaa.com.au/projects_details.php?project=16

[1]http://www.weed-it.com/

Precision agriculture is all about controlling variability. When you have a small plot of all the same soil, under the same weather conditions, your really not going to have much, if any variability, and probably won't actually need any precision agriculture. As opposed to a 400ha paddock where you can have 3 different soil types and different weather at either end of the field.

As for this technology being used on a larger scale, I'd be really interested to see the cost for a similar setup over a few thousand hectares, and what sort of efficiencies might be gained compared to our current system. However I'm not convinced it'll replace our machinery any time soon.

The other interesting use for this would be, if it gained a large enough user base, the sharing of different planting setups and rotations for nutrient replacement. i.e planting crop x and y in rotation, or interrow with crops a and b.

Local farmer groups are already a thing, where farmers get together to share research, look at properties with trials running on them etc. [1]

The question your have to answer before trying to sell any of this "info", is who, other than the farmers themselves is going to find it valuable enough to buy?

[1] http://gga.org.au/group/northern-agri-group/

What I think is going to be most interesting is companies/banks/insurances agencies being able to gleam information without going through the farmers at all. Not by subverting them and going to the machinery manufacturers, which is already happening, but just by using other sources of information, such as satellite imagery.

It's already possible to predict yield potential through biomass/NDVI monitoring via satellite, and given that imagery from a source such as LandSat is freely available, or something like Planet isn't that expensive once your a bank/insurance agency I'm certain they must already be using.

Once you can predict yields reliably via satellite, you can start to predict supply for certain types of grain in different regions, or to a lesser extent, countries as a whole. This data is useful to investment banks and insurers, as well as organisations such as CBH[1], that handle the entirety of Western Australia's grain export. They currently determine potential yield by sending out a grower survey, and asking the farmers to predict yields themselves.

I think the idea of anyone "stealing" our data is overrated. The data from our farm[2] is specific to our farm, and as long as where not farming in such a way that is detrimental to our partners (such as banks/suppliers), such that we were trying to lie about yield potential, or resource usage, I'm not sure what value could be stolen by anyone with our info. In fact we openly share all the trial work we do, so that everyone in the community can learn and benefit.

[1] https://www.cbh.com.au/ [2] https://hectare.ag/nook

Can you link to any of the work you've done, or any articles that explore similar ideas? I'm interested in reading more.

We've been using variable rate to tailor nutrient applications for a few years now[0], and have started contracting with a WeedIt to do spot spraying(major reduction in chemical application).

We've definitely discussed "reforming" some of our paddock structures to improve yield, or mitigate weed potential (sowing east-west vs north south). Although if you were going to commit to major production changes like that, you'd probably want to bundle them together. Changing sowing direction would mean restarting CTF, so you'd be best of doing another operation such as deep ripping at the same time.

We've experimented with wetting agents, as we have quite bad non-wetting soil, and the results have been quite good. It's amazing to watch the before and after[1]

Do you think we're going to need a shift in machinery before we can break out of conventional cropping patterns/methods? I'm still somewhat skeptical that a swarm farm type setup would work in our broad-acre situation, but I think a downsizing of machinery (1 * 60m bar -> 3 * 20m bar), could work if they also move to fully autonomous & electric drive. Although, automation for field operations is still quite "simple" compared to the logistics of managing seed/fert refils.

[0] A few example maps - https://hectare.ag/nooka [1] The effects being discussed here are different, but the video shows the same idea https://twitter.com/ipaddockapps/status/708066704954249216

Could you expand a bit more on what you'd like to see in a whole-systems redesign? I'm currently trying to work out what is going to be the next "evolution" in our farms operation (~10000acres in Western Australia).

In regards to your 4th point, our cropping systems in Australia are a bit different to the states. We already grow crops in a rotation, Wheat, Lupins (Nitrogen fixing) and Canola. Depending on where your farming your might also be growing cover/summer crops.

In Australia, there's a far greater yield benefit to be gained from farmers buying a deep-ripper/mulboard plough(break the hard pan and allow root growth), switching to CTF(prevent a hard pan from happening again), switch to a no/minimum till system, spreading lime(correct soil pH) and maybe spreading clay (mitigate soil erosion).

All of these practises are also "low-tech". You'd want to use VR maps to spread your lime[0], and possibly clay (although it's fair easier to pick out soil erosion by eye as opposed to varying pH), so it's probably not needed.

Once you've sorted out the fundamentals, then it makes sense to start more precision ag. There's no point trying to grow a 5t crop when your roots can't even get down to the soil moisture.

[0] Example VR lime maps, pick Lime from the available layers - https://hectare.ag/nooka

"..have to borrow before every planting" This is not limited to just small farms. In Western Australia, farm debt has been growing 8% per annum since 2000. The average farm size here is ~3500ha (8650acres).

This setup seeds at 2.5 acres a minute. In the video it's being driven by a Baldwin, but has also been run with dual JD track machines.

It uses variable rate and section control; so there shouldn't be any input wastage when the runs overlap.

It also folds up for transporting. https://twitter.com/onus_agronomy/status/732031765204766720

How big do you mean when you say 'huge fields'? Attending to single plants individually, this would be far to slow to work over a huge area.

'In contrast to modern agriculture, where things are done with huge monster tractors' - You might be interested in what http://www.bluerivert.com/ are doing. They are essentially attending to every plant individually in smaller cropping areas.

Once you actually get to a large area, I.e 1000ha fields, it doesn't make sense to attend to every plant individually throughout the season. We do however have technology to plant each seed down to mm precision, and apply nutrients based on zones created within fields.

Also, does it make sense to duplicate all the infrastructure required for one of these, over every single paddock on a farm, when we could have a single, essentially autonomous, machine crop and manage the entire area itself?

I'd be interested in hearing your arguments/use cases for drones over broad-acre properties, we've got(generally wheat) producers here with properties > 150'000 acres.

"can capture geospatial data on huge areas much more easily"

What are you referring to as a "huge" area? All the work I'm seeing done with drones has been limited to ~700ha/day, if you've got perfect conditions. Normally it's much less.

What do you think drones with heat or IR cameras are most applicable too? When we've got machinery mounted sensors such as a biomass/NDVI camera or a weed-seeker, and the machinery has to make a pass over the field anyways, why not have those sensors onboard the machine? We're seeing Weed-seekers pay for themselves in the first year of use.

"People should have the right to tinker with whatever they bought"

Absolutely. We've had so many on farm productivity gains by modifying our machinery. We've added curtains to our spreader to better spread lime, extended our machines axles to setup a CTF system and built our own wetting agent system for our seeder bar, just to name a few, all which required machinery/implement modification. In some cases, machinery manufacturers have come and looked at the modifications we've made to a machine (in the case of the spreader curtains), and then taken the idea and started building it into their machines by default.

Which provider do you propose we boycott, and which one do we use in the mean time? JohnDeere, CASE, New Holland etc dominate the industry, and no farmer is going to (let alone can afford to) stop using their current equipment, and spend millions of dollars (it would easily cost us this much) buying different brand machinery. Let alone dealing with the hassles that come with swapping from one stream of ag software to another.

What's most likely to happen is farmers will just keep modifying their machinery at will. It's already common place to chip your engine, or to physically modify machines/implements to improve performance or your production methods. I don't see that changing anytime soon.

1) Autonomous farm machinery (more than just auto steer) becomes more commonplace.

1a) Release of the first non major brand (JD, Case etc) autonomous machine.

2) Drone see uptake in precision spraying applications. Although this will be on small farms. Broad acre will still be too hard for a while yet.

3) NDVI becomes one of the most commonly used inputs pre seeding and for nutrient applications.

Yes, a week is far to slow. We need to be able to spray multiple paddocks a day, and respond to weather events quickly. I.e you may to want to spray a few thousand hectares in anticipation, or right after a rainfall event.

So our current sprayer covers ~70ha/hr. That means we need a fleet of 18 drones to match it. That's 270k + the associated inferstructure costs. If there not fully autonomous you'll also need to pay someone $30-$40/hr to stand around an continually refill, program and move them from paddock to paddock. As well as ensure there aren't miss sprays, crashes, spray drift onto neighbour paddocks etc. However given the flight time is <20 minutes you'll need far more than 1 person for that, maybe 2 or 3.

We can hire a spray plane for $12/ha, or a contract sprayer (self propelled boom) for slightly less than that.

Our current self propelled sprayer can cover roughly 70ha an hour. 36m boom travelling between 18 and 25km/hr. Water flow rate also plays a part in how much you can cover.

I can see the possible benefit for small vineyards & farms, but at 4ha/hr it would take more than a week to cover a single paddock (field) of ours. Although I'd assume the controlling software would have good route planning, probably driven by drones doing weed spotting and NDVI/biomass analysis

I'd also like to know more about the kind of infrastructure you'd need to support autonomous operations, because if they're not autonomous they still a step behind current farm equipment.

Interesting how the WSJ and BBC articles contradict each other, one claiming plans are inexpensive while the other claiming they are. We use planes to complement our machinery, and they definitely aren't prohibitively expensive, and have far more use than just spraying chemical.

There's a research project being done on the viability of hand picking radish plants out of wheat and lupin crops on broadacre farms here in Western Australia. Some farmers already do it, but once the results come out, I'm wondering home many more will start.

What do you envision having it mounted on that will drive at the speeds mentioned in the article? It's not like you can throw It on the front of your regular dual wheel or track machine and operate efficiently at those speeds.

So you now you have to run your fleet of robot weeders over the field twice? That seems like a waste of time and $ when you could achieve the desired result in a single pass. every extra pass is also more soil compaction.

This is pretty awesome, and I'm looking forward to playing around with it this arvo. Although I somewhat disagree with "Resolution at these bands is 20m per pixel, a lot less than the sub-meter imaginery of Google Maps, but enough for most crop fields."

From a whole farm overview point of view, 20m/pixel might be ok, i.e comparing entire paddock response, but for creating useful maps and applications in the field, we really do need better than that. We can individualize input control down to a per nozzle (plant) basis, so ideally the data we use to create maps should be as close as possible to that resolution.

Every year we take on workmen/students normally from outside Australia, who need to spend time actually working on a farm to finish their agricultural degrees or similar. It's always interesting to see their reactions after they finish telling us about their "big" farms back home, when they realise we can fit their entire farm + change into a single paddock here.

Land may be cheaper, but probably not higher yielding. When your buying land your essentially buying rainfall. We could easily find and buy xx'000 acres of really cheap land, but it if only gets 200-300mm of rain every year, your not going to be seeing serious yields. The other thing to remember is that irrigation is impossible for a lot of West Australian farmland.

Predominately wheat. The rest is lupins and a small amount of canola (~750ha this year).

I'd like to get a more comprehensive look at FarmLogs, hopefully they'll be more available in Australia soon. Although until then, after just browsing through the interface, I'd like to make a few points. I should point out I'm from a broad-acre farm of > 10'000acres, and we have single fields nearly 2.5x bigger than the 600acre size mentioned..

There's still seems to be to much of a focus on farmers having to manually enter data. Our machinery already records everything that happens in the paddock, there's no reason why we should have to sit down and manually enter everything we do. Farmlogs should connect to whichever machinery provider we use's API, and pull all the data straight from the machine. It's now possible to do that in real time, and it opens up far more interesting opportunities, (for example optimising variable rate maps in real time) as we can also push data back to the machines.

How do I connect and use my local weather station data within FarmLogs? One limitation of our current yield prediction software is that it's weather data feed comes from a BOM station nearly 30km away, that isn't nearly good enough, so we're seeing a transition to on farm weather stations. Ultimately we need to be able to collect weather info on a field by field basis, not only for crop modelling and yield prediction, but to better optimise harvesting/swathing schedules, and make more effective N applications (we can create better application maps based on the yield models + predicted weather).

Can I import my all my soil data? We've had extensive EM38 mapping, radio-metrics and soil testing done over the farm to better calibrate our yield models, and improve nutrient application, how do I put that data into FarmLogs? If I can, how do you use it? Can you generate my VR maps based on soil data + predicted weather information?

I'm also seeing some features that most farmers I know will never use. We don't need "another" calendar. We don't need to keep a record of all the grain we have stored in silos, or upload information about every piece of machinery we have on farm. What benefit does that give us? It seems like your almost trying to gameify farming with all the graphics..

I think tools like FarmLogs need to move away from just data collection/consolidation and much more towards yield prediction/modelling, and giving recommendations/optimising nutrient applications. We spend more than $500'000 a year on fert/chemicals, so if you can help me make a more precise variable rate map and save me 10% of that figure, thats something I'm going to use. Even if your charging me $5000/10'000/x'000 a year for it.

In Australia we've seen nothing but farm getting larger over the last 15-20 years. It's also interesting to compare the size of farm he's talking about to what we have here. On our property we have single fields 2x the size of what they consider a small farm (600acres).