Thursday, October 16, 2008

Autofarm: The results

Okay then. Farming requires the raising of plants and animals, for food or textile purposes. It's hard work involving loads of tasks. Let me see if I can relieve a few.

I think we should start with an underground drip system to relieve the task of watering, since plants need water, and farm plants often need more water than is provided by their environment. We dig a system of pipes that drip water up into the soil. Scattered water sensors determine how much to add. The best time to water is 3am, so the plant can absorb it without the heat of the sun evaporating it, or the opportunity for fungus to grow while the plant is not drinking in the evening. Nutrients can also be added to the water.

Now I'll start through the farmer's tasks, automating when I can.

The soil must be prepared, because plants don't grow well in compacted soil. Traditionally, it is tilled, originally by digging it all up, and tossing it back into place. Now tractors do this. I propose an "autotractor" that can plow through a specified area, such as a radio fence (which will provide other benefits later), automatically driving between the points until the entire area is plowed. The farmer can still ride if he wants, but I suspect enjoying a glass of lemonade inside will be done instead. Riding should be done on the debugging stage of this to make sure that it does in fact stay within the radio fence.

The seeds must now be planted. This can be done on the back of the tractor. If computer controlled, it could plant certain seeds in certain areas.

The seeds grow. Other plants also grow, that the farmer wants to discourage. Weeds, in other words. For all the advances in machine vision, the best system for identifying weeds is still the farmer's eye and brain. Still, there have been a few developments in weed eradication over the years. First there was the hoe, which allowed the farmer to dig up weeds without bending over, which was bad for her back. Then there were genetically engineered crops that were immune to an herbicide that would be spread over the field, eliminating everything but the crop. Since the genetically engineered option annoys people, I'll provide some kind of power-hoe at this stage. A number already exist in the market.

If the weeds are gathered, we can soak them in a bucket of water in the sun to extract their nutrients, and add the extract to the water supply. The weeds should then be thrown away or fed to animals.

Once the crops reach a certain height, mulch can be added to retard further weed growth.

The crops must be protected from pests. Symbiotic relationships with the pest's predators is one option, as are drugs that kill the pests. These pests can be animal, fungus, or parasite, and eat the crop mostly because it is there. The radio fence can be set up to make an ultrasonic sound to annoy animal pests away. Sound also benefits the plants by some means not currently understood. Perhaps it knocks air into the plant's stoma.

At harvest time, the plant should be uprooted and the parts of human interest cut from the plant. Mechanical harvesters exist for most crops I could name. Lettuce provides an interesting challenge. I propose scooping it up from below, banging it against a surface to move dirt stuck to the roots, washing, and placing in storage.

Animal care is already quite automated due to immense interest from the meat industry. (Automation means less paying for labor which means cheaper meat.) Animals are kept in specialized pens that feed them, remove their waste, and protect them from fighting with each other. Water is also provided. Chicken coups are designed to separate the chicken waste from eggs, and funnel the eggs to egg packaging departments and the waste to some sort of disposal. (Or on the crops, as chicken waste is actually a very decent fertilizer.) Cow care is also very automated, with auto-milking machines, feed throughs, and so on. A hoof cleaning machine could probably be built with existing sensor technology.

Pigs should be washed more frequently than is practical to do it manually. I propose a system with a touch-sensitive brush that has a supply of water and soap. The brush finds the pig, gently rubs it first with water, then with soap, then more water to remove the soap. After seeking its way over the entire pig's body (minus the face area, I can't think of a way to keep the soap out of the pig's eyes), it can be retracted. The pig will probably enjoy the experience as a gentle massage.

Sheep sheering would be a difficult challenge for sensors. We'd start it by providing the tools under farmer intervention. First the tool guided by the farmer. As the system gets smarter, the farmer will need only watch and be prepared to hit a kill-switch if it goes to far, then ultimately be freed of this task altogether. I imagine the tool would find the sheep, buzz off an inch from rump to neck and the underside, then check the under-material. If wooly, continue shaving. If not, stop and release the sheep.

One last potential invention of interest would be an auto-vet that could monitor the health of kept animals. One problem in large farms is that it is not possible for the farmer to keep track of which animals are sick, and the herd is large enough that at any given time at least one animal is sick. The usual solution of giving all animals antibiotics is unfortunately breeding antibiotic resistant bacteria. The auto-vet could monitor traits of the animal for signs of sickness (waste bacteria count, alertness, sounds made by the animal, visual eye condition, and so on), and should it decide that the animal is sick, provide antibiotics into the animal's water. This would not only reduce antibiotic consumption (which does add up), but also increase the effectiveness of antibiotic doses. Antibiotics would be kept up for a period suggested by a human vet.

Well, it seems that farms were already quite automated before I looked into them, so this article wasn't too worth it. Crap.

No comments:

Related Posts Plugin for WordPress, Blogger...