FarmsReach Groups » FarmsReach

2.5

Lalitha Vish | Jellicles Farm | Sunol , CA

Subject: Soil sampling rover and other farm bots

while working and doing my homework on my small acreage farm bot, i came across an interesting use case for automation in the field.

i spoke to the founder and got some details. It is made for soil core extraction and has a couple of depth/densities option.

85 acres/hour. 800 acres/day. 6 sample cores every 2.5 acres. It even sleeves the soil cores. GPS-RTK. Moves and selects grids. Will remember the spot so it will go back to same spot the next year. Tracked. 5500 lbs. works best in fall and spring.

Mine is a multi functional platform for a smal acreage(sun 100 acre) farm. Would like ideas on what else farmers would like such a roaming/mapping/sensing platform could do....and how many hours it would save in a farm block of 5 acres. The key is to be able to cram as many tasks in one pass. After talking to him, I realized that soil sampling is also a valuable function for a small farm bot. For example.... with strip tillage..ripping the planting area and injecting fertilizer and using a roller does many tasks in one pass. There are many functionalities that can be layered on my bot.

Yield prediction is another. How valuable is yield prediction to a farmer? How much is it worth it to know every week how many tomatoes or strawberries or citrus is ready for harvest. would Farmers pay for this? Would farmers be able to use this data to make money or negotiate better prices etc?

These are some of the things that I have added to my bot... i would like to hear other suggestions

1. Soil sampling

2. pest monitoring

3. vine de-leafing:

4. Mapping and laser leveling

5. measured fertilizer injecting

6. Selective sprayjng

http://7.retrieving samples(leaf, produce)

8. Flame weeding

9. Tree/plant hole digging with an auger

10. orchard spraying(night time)

11. Pest/disease prediction(monitoring outside perimeter for quarantines etc)

12. Yield prediction

13. Seed collection

14. Employee training(using something like google glass but its proving to be more difficult than i imagined)

of course mowing, seeding, but those are broad acre jobs. What else should I add so automation would help farmers? And how much would it save them?

What should be the minimum size? my design is a small tractor sized robot for every 5 acres. That’s the size I thought was optimal given multiple tasks. bigger farms can swarm multiple bots. It would likely cost 30-50k with some basic functions like mobility and mapping(I don’t want to offer robotics as a service.) and other programs can be added or like a subscription or download from the cloud for specific crops like pest details for a specific zone or time of the year. I also want to make it possible to sell data in a farmers market place where there will be buyers wanting farm data. I believe strongly that farm data belongs to the farmer even after produce leaves the farm and that the farmer is owed a share after value is added in the supply chain. That’s a little more tricky but I am working on it.

any input, suggestions and wishlist items? I would most appreciate it as it will help me build a better bot.

Thanks, Lalitha

In Topics Equipment

In General FarmsReach community

Comments 8

02/07/19

2.5

Lalitha Vish | Jellicles Farm | Sunol , CA commented

Thanks, Hasan. That was valuable input. I agree that agrobotics is multi platform and multi disciplinary. There is no doubt that it is tough. Having said that, we ..as farmers..have an opportunity to influence the next paradigm shift in how the upcoming version of farming will shape up..when we went from horses to tractors, it changed everything. And then some more. At some point, farming became a domain that was out of reach of farmers‘ grasp...today conventional practices aid and also harm ag production. there is no kill switch to detect and deter excesses of erosion or tillage or chemical pollution or pest pressure.

There are two arenas. Where a new technology is born and the other is where it Is implemented. The impact of the act of farming the soil continues and is cumulative on soil, environment, water and even markets. And now as Ag changes again, farmers have a chance to speak up and taking

<p>Thanks, Hasan. That was valuable input. I agree that agrobotics is multi platform and multi disciplinary. There is no doubt that it is tough. Having said that, we ..as farmers..have an opportunity to influence the next paradigm shift in how the upcoming version of farming will shape up..when we went from horses to tractors, it changed everything. And then some more. At some point, farming became a domain that was out of reach of farmers‘ grasp...today conventional practices aid and also harm ag production. there is no kill switch to detect and deter excesses of erosion or tillage or chemical pollution or pest pressure.</p>
<p>There are two arenas. Where a new technology is born and the other is where it Is implemented. The impact of the act of farming the soil continues and is cumulative on soil, environment, water and even markets. And now as Ag changes again, farmers have a chance to speak up and taking lessons from what has happened after the previous shift In tractor/chemical driven Ag tech, we can speak up and shape the future  Agtech. We can’t oretend that it isn’t happening. We can’t bemoan that small farmers needs are not addressed. We have to speak up.  </p>
<p>It will  be a data play because data is everything esp when we try to automate but knowing and discerning noise from signal is what Agtech companies need to hear from farmers.  If we dont speak loud and clear,, once more the farmer‘s domain will be influenced by those who don’t actually farm and don’t have a stake in the welfare of farmers and the environmen.</p>
<p>I would like to discuss technical possibilities if you are interested. I am not a technical person but since I have been at this for a while, I can understand almost all aspects of agrobotics. I started out thinking that I could make my own until I realized that it’s is a multi domain problem/solution. I am in the Bay Area and can talk by email or phone. Regards. </p>

02/12/19 8:19 AM

Hasan Bank | None | Sacramento commented

Interesting chain of posts and I've seen earlier ones as well starting from 2017. I would like to add a few pointers here and I hope this will inform some of the folks who have an interest in the AgRobotics domain and let everyone to understand about how easy or hard would be the implementation of robotics and AI into your open-field efforts.

I recently moved to California. Before moving here, I've been working in the domain quite a while and we had an Ag related projects -the name of the project and company aren't that important. We were developing the system end-to-end. In other words, we were developing almost everything including software, firmware, and hardware.

First of all, I should warn everybody that the fancy videos about the intelligent systems aren't there yet. Or cool looking demos are most of the time far from reality. We need at least 3-5 years to have better systems and algorithms to have the technology become a bit more mature -especially robotic hardware implementations- and again the solutions are heavily dependent on the type of crops.

The second point is the folks -especially startups- like to add almost anything AI next to it. I am also starting one and agricultural robotics is in my roadmap. But I would be cautious before releasing anything in that regard. Nowadays, you can find control algorithms called AI because it is sexy and attracts more business for the founders. That said, please look in the startups' credibility -not the investment they collected- by the team's educational background as well as what they published -or done throughout their career. This part is a completely separate discussion and I am sorry to move the subject a bit off-topic. But please be careful. Let's talk a bit more technical:

- ML tools (not AI) -or data-driven algorithms- are quite mature and anything ties to the prediction of the soil moisture, crops growth, etc. would give statistically reliable results.

- That being said, there are tons of developments in sensing domain in parallel including vision, contact, non-contact etc. The sensors always have some limits due to their physical limits such as vision systems would have problems to work at night conditions.

- For the heterogeneous or homogeneous swarm robots, we are not there yet. The robots that you've seen in Amazon robotics or other logistics robotics companies have a quite different setup for their robots. The environment and settings are pretty structured compared to the open-field farming setup. With that, when I was setting the project back in days, we started in the structured and closed environment like a vertical farm. In other words, anything which requires mobility in unstructured and uncertain conditions the robotics domain needs more work to do for functioning fully autonomously.

I also would like to add a few points to

1. Soil sampling (This is relatively easy a ground vehicle is able to do that. There are a lot of dependency to the structure of the field though)

2. pest monitoring (The main issue is identifying and extract the object to reason about the pests' progress in a cluttered environment.)

3. vine de-leafing: (Similar answer like #2)

4. Mapping and laser leveling (This is relatively easy to accomplish. There are various tools and robots doing that from drones to autonomous ground vehicles.)

5. measured fertilizer injecting (The issue is not injection here. The main issue is identifying and extracting the object to spray in a cluttered environment.)

6. Selective sprayjng (The issue is not spraying here. The main issue is identifying and extract the object to spray in a cluttered environment.)

http://7.retrieving samples(leaf, produce)

8. Flame weeding (Just be careful fire and battery operated vehicles may not be that safe. Not sure what do you mean. Flame thrower robot??)

9. Tree/plant hole digging with an auger (There are sample systems which depend on the structure of the environment. I don't think there was a real full-blown, complete, and autonomous solution yet.)

10. orchard spraying(night time) (Same answer as the other spraying solutions. In addition, the execution the tasks at night is not the easiest thing for the vision system.)

11. Pest/disease prediction(monitoring outside perimeter for quarantines etc) (Depends which algorithmic method is followed)

12. Yield prediction (Quite possible)

13. Seed collection

14. Employee training(using something like google glass but its proving to be more difficult than I imagined) [ Actually there are quite a few solutions in different domains in robotics. Compare to the other topics this is the most possible one not only with Google Glass but also other types of equipment such as Hololens or Magic Leap. ]

if anyone interested to talk and learn more about new trends I am eager to discuss more.

<p>Interesting chain of posts and I've seen earlier ones as well starting from 2017. I would like to add a few pointers here and I hope this will inform some of the folks who have an interest in the AgRobotics domain and let everyone to understand about how easy or hard would be the implementation of robotics and AI into your open-field efforts.</p>
<p>I recently moved to California. Before moving here, I've been working in the domain quite a while and we had an Ag related projects -the name of the project and company aren't that important. We were developing the system end-to-end. In other words, we were developing almost everything including software, firmware, and hardware.</p>
<p>First of all, I should warn everybody that the fancy videos about the intelligent systems aren't there yet. Or cool looking demos are most of the time far from reality. We need at least 3-5 years to have better systems and algorithms to have the technology become a bit more mature -especially robotic hardware implementations- and again the solutions are heavily dependent on the type of crops.</p>
<p>The second point is the folks -especially startups- like to add almost anything AI next to it. I am also starting one and agricultural robotics is in my roadmap. But I would be cautious before releasing anything in that regard. Nowadays, you can find control algorithms called AI because it is sexy and attracts more business for the founders. That said, please look in the startups' credibility -not the investment they collected- by the team's educational background as well as what they published -or done throughout their career. This part is a completely separate discussion and I am sorry to move the subject a bit off-topic. But please be careful. Let's talk a bit more technical:</p>
<p>- ML tools (not AI) -or data-driven algorithms- are quite mature and anything ties to the prediction of the soil moisture, crops growth, etc. would give statistically reliable results.</p>
<p>- That being said, there are tons of developments in sensing domain in parallel including vision, contact, non-contact etc. The sensors always have some limits due to their physical limits such as vision systems would have problems to work at night conditions.</p>
<p>- For the heterogeneous or homogeneous swarm robots, we are not there yet. The robots that you've seen in Amazon robotics or other logistics robotics companies have a quite different setup for their robots. The environment and settings are pretty structured compared to the open-field farming setup. With that, when I was setting the project back in days, we started in the structured and closed environment like a vertical farm. In other words, anything which requires mobility in unstructured and uncertain conditions the robotics domain needs more work to do for functioning fully autonomously. </p>
<p>I also would like to add a few points to </p>
<div>1. Soil sampling (This is relatively easy a ground vehicle is able to do that. There are a lot of dependency to the structure of the field though)</div>
<div>2. pest monitoring (The main issue is identifying and extract the object to reason about the pests' progress in a cluttered environment.)</div>
<div>3. vine de-leafing: (Similar answer like #2)</div>
<div>4. Mapping and laser leveling (This is relatively easy to accomplish. There are various tools and robots doing that from drones to autonomous ground vehicles.)</div>
<div>5. measured fertilizer injecting (The issue is not injection here. The main issue is identifying and extracting the object to spray in a cluttered environment.)</div>
<div>6. Selective sprayjng (The issue is not spraying here. The main issue is identifying and extract the object to spray in a cluttered environment.)</div>
<div>
<a target="_blank"  href="http://7.retrieving/" rel="nofollow">http://7.retrieving</a> samples(leaf, produce)</div>
<div>8. Flame weeding (Just be careful fire and battery operated vehicles may not be that safe. Not sure what do you mean. Flame thrower robot??)</div>
<div>9. Tree/plant hole digging with an auger (There are sample systems which depend on the structure of the environment. I don't think there was a real full-blown, complete, and autonomous solution yet.)</div>
<div>10. orchard spraying(night time) (Same answer as the other spraying solutions. In addition, the execution the tasks at night is not the easiest thing for the vision system.)</div>
<div>11. Pest/disease prediction(monitoring outside perimeter for quarantines etc) (Depends which algorithmic method is followed)</div>
<div>12. Yield prediction (Quite possible)</div>
<div>13. Seed collection</div>
<div>14. Employee training(using something like google glass but its proving to be more difficult than I imagined) [ Actually there are quite a few solutions in different domains in robotics. Compare to the other topics this is the most possible one not only with Google Glass but also other types of equipment such as Hololens or Magic Leap. ]</div>
<p>if anyone interested to talk and learn more about new trends I am eager to discuss more. </p>
<div>
</div>

02/12/19 2:23 AM

2.5

Lalitha Vish | Jellicles Farm | Sunol , CA commented

True. They do more than that. There is a lot of AI and ML involved(I saw the bot ..when they started out a year ago and their latest version. It’s massive and pretty impressive).

I have a monthly farm robotics meetup or hangout. During farm peak time, I try to send out at least a newsletter type of thing with updates and happenings in Agtech world. let me know if you are interested and I can include you in future communications.

02/11/19 12:27 PM

220

Paul Underhill | Terra Firma Farm | Winters, CA commented

I just found out about Farmwise. I think there is a huge market for their robots up here in the Sacto Valley but right now they are only operating on the coast. I contacted them to see if/when they are going to expand their services. I think weed control is probably one of the best markets for robotics in ag in California. Another will be orchard spraying and mowing. I would guess most almond orchards will be tended by robots within 10-15 years. It will be a huge market.

02/11/19 10:50 AM

Cathy Bartolic | Perennial Petals | Enfield, CT commented

Please take me off this email list...thank you.C. Bartolic 20 20

02/10/19 2:38 PM

2.5

Lalitha Vish | Jellicles Farm | Sunol , CA commented

Thanks, Paul. What did you think of Farmwise? How many acres have they covered for you? Does it help you with operating costs?

02/10/19 1:30 PM

220

Paul Underhill | Terra Firma Farm | Winters, CA commented

https://farmwise.io/

02/08/19 6:30 PM

2.5

Lalitha Vish | Jellicles Farm | Sunol , CA commented

http://www.agrointelli.com/ : I have also been talking to this Danish company. They are ready for roll out and might be interested in CA market. Please feel free to message me too.

02/07/19 1:23 PM

You must be logged in to post a comment. Log in or Join FarmsReach.

Conversations

Suggest a New Category