Robotic Following Cart to Replace Light Duty Tractor Tasks?

[MossRoad]

Here's a good read on apple picking...

Inside The Life Of An Apple Picker : The Salt : NPR

From the article, the average full container weighs about 1000#. The pickers get paid by the box. They pick not less than 12 boxes per day!

So 12,000 pounds of apples per person per day.

The auto cart with a capacity of 250 pounds would have to be able to make 48 trips per day to keep up with one picker.

Just some more stuff to think about.

Maybe apples isn't the best example... oranges? :laughing: Just kidding. Anyhow, we need to find a better example to justify the cost of the device.

 

[MikeFarm]

Hi candersen10

Just came across this thread. Haven't read all the posts but the cart that can follow you around would be so helpful for older farmers that have problems lifting/pushing a wheel barrow. Ace device and I hope it succeeds well.

Mike

 

[Dadnatron]

The serious lawn crews here are running a compact skid steers.

I wouldn't let a lawn crew run a skid loader over my lawn to spread mulch in the spring. But, a small tracked robot might not be too bad. It would depend on its impact on an established lawn.

 

[aczlan]

Here's a good read on apple picking...
Inside The Life Of An Apple Picker : The Salt : NPR
From the article, the average full container weighs about 1000#. The pickers get paid by the box. They pick not less than 12 boxes per day!
So 12,000 pounds of apples per person per day.

1000# is pretty close, we see 750# to 950# of apples per bin (the bin adds another 90# to 165# depending if its wood or plastic and the bin style).

I wouldn't let a lawn crew run a skid loader over my lawn to spread mulch in the spring. But, a small tracked robot might not be too bad. It would depend on its impact on an established lawn.

A large local outfit has a self unloading trailer (like: Products - SS8 Mulch Spreader - Littau Harvester but self powered, it has a unloading chain on the bottom like a manure spreader and a conveyor belt going across the front to load a wheelbarrow). It will load a wheelbarrow in under a minute. 3 guys running wheelbarrows from that to the bed and one guy raking the mulch out will get a lot of mulch spread in a hurry.

Aaron Z

 

[candersen10]

I've been thinking about this quite a bit pretty much all day off and on.... In all seriousness, if the $11K machine can outwork the $4500 machine 4:1 in the first trip, and make 10:1 trip ratio( I think it will be much higher than that), that's outworking it 40:1. If labor $$$ savings is the goal, put it to a spreadsheet. It doesn't pan out. You'd need many, many of those $4500 machines to keep up with one $11,000 machine (really $11,450 with the large light material bucket, plus shipping).

One person picking fruit as an example...

Let's say they can pick fruit and fill a cart sized container in 6 minutes. That's 10 containers per hour. They dump it in the box on the auto cart. It leaves to make the 1/4 mile round trip to the barn and back. The picker continues to pick while its gone. It comes back, he dumps his box on it, it takes off again, etc...

How fast does the auto cart travel? Lets say 3mph, as that's an average person's walking speed. There's 12 quarter miles in three miles, so it can make say 10 trips if you account for dump and load times. That works out well for our example, as the container on it is sized to keep up with one fruit picker person (if the distance averages out to 1/4 mile trips. Short distance works out better, longer distances, not so good).

As mentioned, it can only work for 5 miles. It'll be dead in 20 trips. That's about two hours of work. You'd need 4 auto carts to keep up with one person picking fruit for 8 hours, or 4 battery packs, or a quick charger that can charge it before the shift is over, etc... If you need 4, that's $18,000 for auto carts to keep up with one picker that gets paid only $300 per week working 40 hours for $7.50 an hour.

The picker could load crates in the field and deliver them to the barn himself much faster and cheaper with a tractor. Even if you had enough work to keep the picker busy for 4-5 weeks, that's only $1500 at most. Even if you had continuous crops and enough work to keep a guy busy all summer/fall, its not going to save enough labor to pay for itself because its small and slow.

And another thing to consider. Let's say the guy walks out from the barn to the first tree and the auto cart learns the route. It leaves and comes back to where? The guy has moved to another tree. He'll have to go and get it each time it comes back. That's the same distance spent walking as it would be to a crate set in the field every few trees. So no labor saving there at all.

Look, I'm being hyper critical of course. Just giving you some things to think about. I hope you can make a go of it. It sounds really interesting and poses a whole slew of problems to resolve. Keep working on it and keep updating us and good luck. :thumbsup:

MossRoad, below are the unit economics I was thinking. Apples might be a bad use case, but in certain crops, if we assume the average worker can pick 500 pounds of produce an hour, and then has to travel 500 feet to a collection point over a 10 hour day, that equates to 76 500 foot trips over a 10 hour day, or about 143 minutes walking per day. Assuming 30 days picking per year, that's around $1,080 dollars in labor.

Are there any crops where those types of economics exist (thinking Lettuce, grapes, other items where there is a lot of picking and then shuttling stuff, but also not a ton of weight concentration like apples where you may thus go to big material handling equipment as mentioned by posters previously)? Thinking things like this perhaps:

 

[4570Man]

I wouldn't let a lawn crew run a skid loader over my lawn to spread mulch in the spring. But, a small tracked robot might not be too bad. It would depend on its impact on an established lawn.

Here's what they use and grass fairs pretty well. Even if they couldn't use it on grass most of their commercial properties have asphalt to run them on. A few sheets of $7 OSB board would solve any turf damage problems.

 

[quicksandfarmer]

There are a lot of jobs where much of the job is materials handling, and anything that speeds the handling makes the people doing the job more productive. We've made great strides in the past 150 years or so at replacing the brute force in moving stuff around with machine power rather than human power, but people still spend a lot of time doing it. People are needed to guide the machines where to put stuff, and provide the brute force in places that machines can't go -- like carrying a bathtub up three flights of stairs.

So no doubt that there is a huge market for smarter material handling machines -- machines that can operate without human guidance, or can go places that other machines can't go.
Here's an example of a company that is using robots to move plants around in a greenhouse:
Metrolina Greenhouses: Spacing plants with Harvest Automation HV-1 robots - YouTube
A lot of human labor is being saved there.

Now, autonomous navigation is a hard problem in robotics. Right now, it's the hard problem in robotics. Every major automaker and several other large companies have large teams of smart people working on it-- for self-driving cars -- and whoever ends up solving it is going to be rich and famous. If you solve the problem of autonomous navigation, building general-purpose material-handling robots is straightforward. Conversely, without autonomous navigation -- if a person has to guide it -- a robot is just another material-handling device. And most people prefer a material-handling device that they can ride to one they have to walk with!

Nothing in your video indicates that you've done anything toward solving the problem of autonomous navigation. Following a target visually is not a hard problem -- my kids were doing that with Lego Mindstorms when they were in elementary school a decade ago. And it's not really an evolutionary step on the road to autonomous navigation.

Take a look at that greenhouse video. One of the things that makes navigation an easier problem is to have a controlled environment. If you are working in a greenhouse, where there are walls to limit your movement, the floor is concrete, and you can put navigation aids all over the place, it's much easier to get around than outdoors.

So I would work on navigation first. Start with finding controlled environments. Branch out from there.

 

[candersen10]

There are a lot of jobs where much of the job is materials handling, and anything that speeds the handling makes the people doing the job more productive. We've made great strides in the past 150 years or so at replacing the brute force in moving stuff around with machine power rather than human power, but people still spend a lot of time doing it. People are needed to guide the machines where to put stuff, and provide the brute force in places that machines can't go -- like carrying a bathtub up three flights of stairs.

So no doubt that there is a huge market for smarter material handling machines -- machines that can operate without human guidance, or can go places that other machines can't go.
Here's an example of a company that is using robots to move plants around in a greenhouse:
Metrolina Greenhouses: Spacing plants with Harvest Automation HV-1 robots - YouTube
A lot of human labor is being saved there.

Now, autonomous navigation is a hard problem in robotics. Right now, it's the hard problem in robotics. Every major automaker and several other large companies have large teams of smart people working on it-- for self-driving cars -- and whoever ends up solving it is going to be rich and famous. If you solve the problem of autonomous navigation, building general-purpose material-handling robots is straightforward. Conversely, without autonomous navigation -- if a person has to guide it -- a robot is just another material-handling device. And most people prefer a material-handling device that they can ride to one they have to walk with!

Nothing in your video indicates that you've done anything toward solving the problem of autonomous navigation. Following a target visually is not a hard problem -- my kids were doing that with Lego Mindstorms when they were in elementary school a decade ago. And it's not really an evolutionary step on the road to autonomous navigation.

Take a look at that greenhouse video. One of the things that makes navigation an easier problem is to have a controlled environment. If you are working in a greenhouse, where there are walls to limit your movement, the floor is concrete, and you can put navigation aids all over the place, it's much easier to get around than outdoors.

So I would work on navigation first. Start with finding controlled environments. Branch out from there.

Quicksandfarmer, thank you for the reply. The coding for the prototype you see was done in a day. We know it is not a fully developed product; the question we are trying to answer is not can we build it, but rather, would anyone buy it if we built it and it functioned perfectly as described?

The application I am imagining is walking up and down rows of stuff (grapes, lettuce, tomatoes, mushrooms, etc.) picking into a plastic bin, and then carrying that plastic bin back to a truck or tractor, which is then shuttling it back down to a collection point. With this concept, a picker turns on the Burro (potential name), hits a "home set" button, then walks up to pick; the Burro follows behind him/her with a few plastic bins. When all but the last bin are full, the worker removes the last bin to keep picking into it, and hits a button on the Burro which then retraces its path back home using odometry/GPS, and is emptied by someone packing produce at the collection point. Once emptied, that person hits a "return to Picker" button, and the Burro returns back to the picker ready to be filled again.

The technology is a tricky part; the commercial example you used is an example (I believe) of guys who developed technology for a problem that didn't really exist in a widespread area, and at a huge cost (about $30M raised). Thus, reducing a paint point at a reasonable cost is the biggest challenge. Is this the same type of thing (not a big pain point)? Or do we have a pain point?

 

[quicksandfarmer]

OK, normally you start with a need an design a solution for it, but sometimes it works to start with the solution and then find the problem it solves.

Think about jobs that have to be done by hand, because they are complicated. Then think about jobs that are typically done by an expert and a helper. Why does the expert have a helper instead of another expert? Because the job has a component of unskilled labor and there's no point in paying an expert for unskilled labor. What you are looking to do is replace the helper. Think of your target as what a 16-year-old boy would have to do to be helpful. If all he did was follow me around I'd send him home at lunchtime. A lot of helper jobs fall into one of two categories, either "keep feeding me materials as fast as I can use them" or "pick up what I create and carry it away for collection." Think of a bricklayer, he's got a helper whose job it is to give him a steady supply of bricks and mortar. That's the kind of job your machine has to be able to do.

You want to be looking at places where high-skill labor is used, where you can make someone with expensive skills more productive. I don't think harvesting agricultural products is going to be there, that's all about using the cheapest labor possible.

 

[candersen10]

OK, normally you start with a need an design a solution for it, but sometimes it works to start with the solution and then find the problem it solves.

Think about jobs that have to be done by hand, because they are complicated. Then think about jobs that are typically done by an expert and a helper. Why does the expert have a helper instead of another expert? Because the job has a component of unskilled labor and there's no point in paying an expert for unskilled labor. What you are looking to do is replace the helper. Think of your target as what a 16-year-old boy would have to do to be helpful. If all he did was follow me around I'd send him home at lunchtime. A lot of helper jobs fall into one of two categories, either "keep feeding me materials as fast as I can use them" or "pick up what I create and carry it away for collection." Think of a bricklayer, he's got a helper whose job it is to give him a steady supply of bricks and mortar. That's the kind of job your machine has to be able to do.

You want to be looking at places where high-skill labor is used, where you can make someone with expensive skills more productive. I don't think harvesting agricultural products is going to be there, that's all about using the cheapest labor possible.

Do you produce/harvest many vegetable/fruit products and not see a need then because the labor cost is so cheap to harvest?