The Economics of Robotics

The Economics of Robotics

Postby John Payne on 06 Oct 2010, 17:00

The following, reposted from another forum, was merely the inspiration for this topic. It isn't meant to constrain the breadth of discussion!

"I'd think that electronic noses might eventually be both more sensitive
and more specific than even the best natural noses, but in the near term
we can expect household robots to warn us about gas concentrations that
are becoming worrisome (before they become dangerous), refrigerant leakage,
ozone (indicating spark gaps in wiring or worn brushes in motors), low
O2 or high CO2, etc. Insurance companies might even subsidize their
purchase by lowering your premiums if you get one."
John Payne
 
Posts: 17
Joined: 24 Apr 2010, 18:46
Location: Boulder, Colorado

Re: The Economics of Robotics

Postby John Payne on 11 Oct 2010, 02:56

From where I sit, it looks as though the majority of participants in the development of robotics, industrial applications aside, are in academia, and that the majority of funding comes from government (largely military) sources. In my humble opinion, this is a situation ripe for change.

The list of players in this drama include the usual suspects, the holders of military and other governmental purse strings, educational and research institutions heavily dependent upon such funding, and of course the companies involved in producing the computing hardware that amplifies the value of sensors and actuators, but it also includes companies with long experience in developing and marketing industrial robots, newer companies involved in other aspects of robotics, established companies in markets likely to be heavily impacted by robotics (agriculture), venture capitalists looking to get on board early with enough resources to turn good ideas into winners, and more conservative investors simply looking for a decent return.

On the consumer side, it includes insurance companies which recognize a self-interest in encouraging the proliferation of certain categories of robots, those that can help reduce their pay-outs, transit agencies that decide to provide the infrastructure for automated public transit (PRT and GRT) and/or for private vehicles with an automatic mode allowing them to operate on public guideways, and anyone with the need for 24/7 security. Perhaps more importantly, it includes a value-conscious public which has become bored with PCs and is looking for something both more interesting and more practical.

A critically important aspect of robotics is that it possesses the potential for self-replication, more likely at the factory level than at the level of the individual machine. Whoever owns the first functional, self-replicating robotic factory will stand to reap a windfall.

Given sufficient economies of scale and process, there's a virtually unlimited market for machines to accomplish tasks currently considered uneconomic, because we don't feel we can afford to hire people to do them. Undoubtedly, there will be some jobs lost to robots, but I believe the potential exists for the impact of robotics on employment to be positive, opening more new positions which are less menial than those taken away.

The effect of robotics may prove so profound that it forces a rethinking of the fundamentals of economics. Scarcity (of labor and of necessary commodities) may cease to be relevant, and need a replacement paradigm. Quality is a consideration sure to gain importance.
John Payne
 
Posts: 17
Joined: 24 Apr 2010, 18:46
Location: Boulder, Colorado

Re: The Economics of Robotics

Postby John Payne on 30 Nov 2010, 19:01

So why should a general improvement in quality result from the proliferation of robots?

Quality results from the careful, attentive application of knowledge and experience.

For a baker, this means choosing ingredients that themselves exhibit quality, combining them in proportions that are enough, but not too much, applying processes (mixing, kneading, proofing, baking), observing the result at each step, and using that to adjust the ingredients, proportions, and processes.

Programming and/or entraining a machine to mirror these skills would be no small undertaking, but once accomplished it can be transferred to as many similar machines as you like, which can then share new experience gained through observing the results of many little experiments, filling in the unknown spaces in a multidimensional map of the territory, and either corroborating earlier data or calling it into question - all as part of their normal operation. In this way, initially acceptable results become better and better.

Of course, the machines would have to be taught what "better" meant, about which even experts disagree.

But, in contrast to humans, whose capacity for attention is constrained by biology, for robots attention is a matter of the variety, sensitivity, and resolution of their sensors, the speed of their processors, and the algorithms applied to making sense of sensory data, all of which tend to improve with time. Even with no change in their manipulation hardware, there's plenty of room for improving performance.

Hardware also improves, of course, introducing new sensory modes, mobility, strength, speed, degrees of freedom, etc. But generally the computing hardware improves faster, meaning that over time more computing operations are available for each action taken. This might be very loosely quantified in terms of processor operations performed per erg of physical energy expended.

In practical terms, this means that over time robots gain not only the ability to carry more and move faster, but to do so with ever greater care. A gardening bot, for example, might go from slowly moving leaves aside to get a peek at what they'd been hiding to quickly snaking an appendage between them with a minimum of disturbance. For another example, a baker bot, prepare to grind some flour, might go from scanning the grain in a shallow pan and picking out the inferior bits to scanning a stream of grain as it fell from a funnel, deftly flicking the shriveled kernels aside.

Having put the most thought into this in the context of horticulture, I'll continue with that example. As initially deployed, a gardening bot might have just time enough to perform predetermined planting, weeding, and harvest operations over an acre or two. Early on it might gain the ability to recognize signs of common problems, like nutrient deficiencies or aphid infestations, then the ability to make spot applications of supplements and to remove leaves with high concentrations of aphids, working these actions into its routines. Later it might gain the ability to recognize other pests and diseases, and add appropriate remedies to its repertoire of actions. Eventually, it might gain a sophisticated understanding of ecology, and the ability to act subtly, as by refraining from killing off every last tomato worm so as to avoid also killing off all of the parasitic wasps that keep the tomato worm population under control. And somewhere along the way it might also gain the ability to distinguish rare and endangered plants from weeds, and to treat them very differently. This whole metamorphosis could happen with little change to the basic hardware, although it's more likely that some retrofitting would happen.

With the capacity for care, attention, and knowledge from shared experience all increasing, how could quality not also benefit?
John Payne
 
Posts: 17
Joined: 24 Apr 2010, 18:46
Location: Boulder, Colorado


Return to Podcast

Who is online

Users browsing this forum: No registered users and 9 guests

cron