no one's going to take your tractor away

Even if, between now and January, Congress were to get religion with regard to the benefits to be gained from applying robotics to the transformation of agriculture, and had a full-speed-ahead bill ready for the signature of our new President on the day he takes office, it would still take time for the effect to become evident on the landscape.

To begin with, while you can see the potential for it in what exists today, the technology largely remains to be developed, so figure five years of R&D and experimental installations before anything starts rolling off an assembly line, and probably another five to get the bugs out to the point where it's really possible to let the machines run without some degree of supervision.

At that point, ten years hence, you might still have to drive twenty miles to see one of the new machines in operation. Then, for at least another ten the story would be one of them becoming very gradually more common, as well as more sophisticated. Meanwhile there's mouths to feed, hundreds of millions of them, and business as usual will necessarily continue.

At some point, maybe twenty-five or thirty years out, the size of the market for food cultivated by autonomous machines would surpass the size of the market for conventionally grown food, and at about that time I would expect to see several things happen. For one thing, the largest tractors would disappear from the market, as there would no longer be sufficient demand to justify their production. Also, a shakeout would begin among tractor and implement companies that hadn't gotten into robotics themselves. On the other hand, the infrastructure for getting grain and produce to market could be expected to improve, under pressure from robotic operations with their more diverse output and their ability to provide detailed information about what they would need to move how soon.

Some crops, however, would continue to be more economically produced by conventional methods. In particular, it would be difficult for generalized machines using horticultural methods to match the efficiency of traction-based monoculture in the production of small grains - wheat, barley, oats, and rye. At least in the near term, the reduction in acreage dedicated to raising these crops by conventional methods would result not from direct robotic competition, but from the substitution of more fruits and vegetables in place of grains in the diets of both people and livestock.

Granted, with the advances in robotics that all of this activity would bring about, the tractors on the market then would likely also be capable of autonomous operation, although I'd expect to see a lot of hold-out farmers, using older equipment, still spending long hours driving their tractors, powered by synthetic fuels. Whether that practice will ever become as uncommon as farming with horses had become in 1960 is anyone's guess.

the importance of interfaces

Take the USB port as an example. It's ubiquitous; practically everything either has one or plugs into one.

Similarly, if you want to build a multi-vendor market for almost anything, one of the best things you can do is to find the natural divisions of responsibility and insert standard interfaces into the boundaries between them.

One example, in the context of cultibotics, would be the connections between robotic arms and tool units that attach to them. What physical form should the connections take? How much force should the mechanical connection be able to withstand or apply? What services should the unit be able to expect from the arm? What signals should each understand or send to the other, or pass through to the CPU? Would the arm supply water, or should any unit making use of it have a hose connected to it in addition to its connection to the arm?

Detailed answers to these questions would fill a thick book, which is what it typically takes to specify a standard. Moreover, chances that any standard organization which undertook to fill in the details would decide that there would need to be several such standards, to accommodate scales ranging from very small to very large.

But given a set of well-defined standards, you'd be able to buy a robotic arm from company A and a tool unit complying to the same standard from company B, and have good reason for confidence that you could just plug them together and have it work seamlessly.

Until now there hasn't been much need for standardization in agriculture. The prime examples of what there has been would be power takeoff, hitches, and hydraulic connectors, all of which have been standardized by the ISO, which makes it the most likely candidate for tackling standards for robotics in agriculture.

Of course the ISO isn't going to get involved until there's at least the beginnings of a market and more activity than sparse experimentation, so it behooves those who do get involved early to cooperate with each other to develop ad hoc standards which are in the public domain, royalty-free, or available for low-cost-per-unit licensing, suitable to the bootstrap nature of the field. These ad hoc standards can later serve as the starting point for formal standards.

finding a place for cultibotics in Obama's rural agenda

Essentially the same post also appears on my general purpose blog.

It's not like there was any shortage of ideas for how to improve the stability of U.S. agriculture, the lot of farmers, and the economic vitality of rural America. Just have a look at President-Elect Obama's rural agenda.

What would the ideas encapsulated here look like if embraced by the Obama-Biden team? What might they be called? Here's a few focused statements that occur to me...

• Help insulate farmers and farming regions from dependency on volatile bulk commodity markets by encouraging greater diversity of production.
• Facilitate production improvements through both simultaneous and sequential polyculture.
• Enable farmers to grow more of their own food without need for much time investment or manual labor.
• Reduce the time spent in machine operation.
• Reduce the acreage needed for an economically viable farming operation.
• Reduce the initial investment required to start a farm.
• Provide farmers and their children with high-tech experience.
• Create a demand for skilled technicians and technical instructors in rural areas.
• Create opportunities for rural youth.
• Preserve local crop varieties and experiment with new crops.
• Improve the quality and diversity of locally available produce.
• Reverse the impoverishment of rural culture.
• Reduce exposure to pesticides and pesticide residues.
• Reduce the dependency of agriculture on fossil fuels and feed stocks.
• Reduce contamination of runoff and ground water.
• Reduce and eventually reverse the loss of soil fertility.
• Reduce wind-borne dust.
• Enlist productive land in the efforts to preserve endangered species and provide wildlife habitat.

This list could be far longer, but that should be enough for a sample.

more specifically, how would they work?

This a subject for research and development, of course, but it's my ‘job’ to make this vision as accessible as I can, to both anticipate what that R&D might produce and describe it in plain language.

First, these machines will necessarily have sensory components. Digital cameras and microphones are practically a given, but they may also have infrared imaging, radar and/or laser scanning, chemical sensors to provide something akin to a sense of smell, pressure/stress sensors for a sense of touch, probes for soil moisture, temperature, pH, O2 content, and nutrient availability, weather instruments, and some means of locating themselves very precisely relative to the boundaries of a field or other stationary reference. Compared to most machines, they will have available a rich collection of information about their environments, rich compared even with what human senses provide.

Next, they will have significant computer processing power, sufficient to take the data streams from all of these sensory devices, find patterns in them, compare them with each other and with historical data (including the exact position of every seed and when it was planted), create and update a real time 3-dimensional model of their immediate surroundings, locate items of interest within that model, choose a course of action, and send the detailed instructions to the machine's moving parts, closely monitoring their progress.

Finally, they will have various moving parts, likely including high resolution or specialized sensory components that can be sent in for a closer look. Those moving parts might include a range of grips, from fine tweezers to something strong enough to uproot small trees, mechanical snips, lasers with enough power to fry a meristem, high-pressure water jets capable of slicing through the stem of a plant, fingers to move other plant material out of the way, a vacuum for sampling air at ground level or removing insects, sprinklers and sprayers, trowels of various sizes, and, of course, the soil probes mentioned earlier. Such tools might be combined into sets incorporated into units which could be plugged onto the ends of articulated arms and quickly switched out.

That's a basic outline, but we need to return to the data processing hardware and the code it runs to fill out the picture, since it can make the difference between an expensive toy and a productive machine that more than pays for itself. A major task the processor must perform is resource scheduling, and to do that effectively it must sort actions into those that can be performed without moving anything massive (slow) and without switching out tool units, those which require either movement or a tool switch but must nevertheless be accomplished before moving on, those which can be left until a future pass over the same area but not indefinitely, and those which can be left undone unless it becomes convenient to do them. Efficient scheduling also means mapping the movement of even the smallest parts so they proceed smoothly from one thing to the next, without having to retrace their paths more than is unavoidable.

An important point to be taken away from the previous paragraph is that scrimping on computing hardware and software is likely to prove counterproductive, by reducing the overall capacity of the machine disproportionately. We should expect the computing components to represent a substantial fraction of the overall cost of the machine, and we shouldn't be surprised if they also consume a substantial fraction of its energy budget. Better to invest an extra 10-20% to make a given physical machine capable of performing the work of two, and to invest 1 or 2 kilowatt-hours to save ten.

Something which should be apparent from this mental exercise as a whole is that what's being proposed is largely a simple extrapolation of technologies which already exist. There are already mechanical arms and mechanical grips; there are already sensors and various means of controlling machine operation. What's mainly missing is the software which would turn data streams into a 3-d model in a horticultural context, choose what to do, schedule resources, and map out the details. That's a lot left to be done, requiring a significant investment for a long term payoff, but it's a fairly straightforward problem, and divisible into more manageable chunks. Let's get to it!

imagine a machine built for efficient gardening

What would it look like? How would it be powered, and how would it transmit power to the parts that need it? What actions would it be capable of performing?

There's no single, right answer to these questions. Rather there's a wide range of potential answers, some of which will likely prove more workable than others. Let's look at some of the possibilities.

What would it look like? Almost anything, from a snake-like device slithering along the surface, to what appears to be little more than a single wheel rolling about, to a platform supported by long, spider-like legs, to a beam supported at each end by wheeled trucks. It may turn out that the best arrangement is a mixture of larger and smaller machines, with the larger ones designed to never put their weight on soil being cultivated.

How would it be powered, and how would it transmit power to the parts that need it? They could get their power directly from the grid, from engine-driven generators, from wind generators, from photovoltaic panels, from concentrating solar collectors, or simply from batteries or other energy storage. Any such machine will need at least a small amount of electricity, to power the electronics. Mechanical power could also be electrical, but needn't be. It might be provided via compressed air. Delicate, articulated parts might be moved via fine cables or wires, much as our own fingers are moved by tendons linking to muscles in our forearms.

What actions would it be capable of performing? Planting seeds, of course, beyond that the possibilities are nearly endless, but even the placement of seeds can be accomplished in many ways.

In conventional agriculture, seeds are typically inserted into the soil in rows, through an opening created by a disk (a rotary knife), and covered over by a roller. This is an efficient method of planting a large area to the same crop, or even to a mixture of crops with seeds of approximately the same size, if you don't mind running the planting device and the tractor pulling it over the same soil surface through which the seeds will have to sprout and in which the sprouts will have to grow. Most such planting devices require the soil first be prepared into a seedbed, meaning that plant debris from previous crops must either be turned under, with a plow, or broken down by a combination of tillage, decay, and weathering, to form a relatively uniform surface, easily broken into small particles. A few such planting devices are capable of placing seeds through rough plant debris, rendering the preparation of a seedbed unnecessary, but they're still mainly used to sow a single crop to a large area.

A robotic gardener would also need to be able to place seeds not only through plant debris but between standing plants. It would do so one at a time, perhaps very rapidly, like a sewing machine, but still one at a time, and far more precisely than any bulk planter, positioning them in the most advantageous microenvironments available. Even when planting in bulk, the use of rows would be optional, and in many cases a honeycomb-like pattern might prove preferable.

Weeding might be accomplished by identifying weed seedlings and removing them while still in the sprout stage. Undesirable plants that propagate by root spreading could be controlled by injecting steam below the surface, wherever they appeared. Insects, like aphids, could be controlled by removing infested leaves. Diseases could be controlled by removing infected plants. Nutrient deficiencies could be identified early and treated quickly. If plant debris needed to be reduced to mulch it could be clipped off at ground level and shredded, without disturbing the soil.

In each case the action taken would be local and specific, rather than applied to an entire field, and generally would not involve moving large amounts of soil around, a practice which wastes both energy and soil fertility.

But the essential requirement, without which this whole scenario would be futile and meaningless, is that the machines must operate autonomously, puttering through their days without constant human supervision. They must have both the ability and the latitude to choose what to do next for themselves. Considering they must also operate in uncontrolled environments, this is the greatest challenge.

panning back to the big picture

This blog is about the application of robotic technology (machine intelligence combined with sensory input and operational flexibility) to the performance of horticultural techniques on an agricultural scale, without continuous human supervision, and with relatively low power requirements.

It is about machines bringing to bear not only cropping plans but, eventually, an understanding of plant ecology in choosing their detailed actions, balancing the need for production with other concerns, including the preservation of endangered species.

There is no example I can point to, because only bits and pieces of the technology currently exist.

My role is to point to what could be, what will be if we set our minds to it.