Developing a robot worth buying

Many (maybe even most) "robots" one comes across tend to be built with the primary purpose of obtaining a degree (MS, PhD, etc). Once they've served that purpose, development tends to stop.

I think that to make a great product, you have to stop thinking of it as a "robot", but instead think about what problem needs solved - what the potential customer wants. One owner of a drill company was purported to tell his sales force - "We don't sell drills, we sell holes." Stop getting all caught up in the bells and whistles of your product that no one really cares about and get inside the head of your customer and solve their problem. Then you have something that people must have.

Roomba sells not because it is a robot vacuum cleaner, but because it's a hands off vacuum cleaner that happens to be a robot. We have one and, contrary to some reports, it works great. No, it doesn't compare to our Dyson Animal, but turn it loose in the family room and it does an adequate job, all by itself, and finds the charger and turns itself off when it's done. It's effective in that we just bring out the Dyson less frequently for deep cleaning.

The really great "robots" may not be considered robots at all by many. Dishwashers and clothes washers. But the idea is the same - it's a machine that saves time and effort so you can do something else that you'd rather be doing. That's the killer app for automation.

But for purveyers of automation, it's really hard to compete with human labor as MLW's $400k security guard can attest. How about a robotic grocery shelf stocking robot - a potentially huge market to stock the shelves of stores around the world? But there's just no way a machine can compete with a high school kid working part time for a few dollars an hour.

The bane of automation is lack of environmental controls - which is pretty much the definition of the "real world."

So perhaps most "robots" are built to satisfy the creative outlet of tinkerers. Or to produce a degree. Or both.

Yep. And where it's not saturated, the killer app just hasn't yet been invented.

What we need are some practical, low cost super conductors. Something as easy to buy and use as a spool of regular copper wire. That and about a 10-fold improvement in battery/power technology. What a boon to invention / engineering that would be.

-Brian

Just some random thoughts on the subject...

I like gadgets. I like things that do "cool" or interesting things that entertain me. I like things that can be change to do something else, either be design or by experiment. I like things that teach me, that help me learn. I like things that can be used in new ways. In short, I like things that can evolve with my interests. I pay a lot of money for my "toys" that fit this need; PS2 games, PDA, computers, robotic projects. I think that to some degree this is all true of EVERY robot builder. I think this need for entertainment that can change with us is the underlying reason most of us are interested in robots.

That being the case, your product not only needs to be interesting or useful right out of the box, but it needs to be designed with the idea that the customer will want to upgrade, change, or modify it from your original design. The market is full of fixed-use robotic products (motor controllers, h-bridges, robotis toys and "kits" that have no flexibility, etc, etc) and it's unlikely you'll have much success trying to sell another one. If you want to produce a sucessful product, make it flexible, or at least hackable. As an example, look at the RoboSapien. It's just another robotic toy, but it was designed right from the begining with the idea that it could/would be "hacked" by the customer. This one idea is probably what has made it so sucessful where so many other toys were not. As another example, look at the Lego Mindstorm kits. It's a nothing more than plastic pieces that fit together, a couple of incredibly simple sensors, and a controller "brick"; yet it's made $Millions.

I guess my point is, regardless of what you make, design it so that it can be changed from YOUR orginal design into OUR unique designs.

WEC

I am thinking that the solution for someone who finds stamp basic too complicated and mindstorm language too limiting is to use Logo.

If you find time to look at it, let me know what you think of it as a basis for robot control. (Of course there is no need to pick just one; I can see merit to providing several control languages).

(Repost of an article posted in comp.robotics.misc)

From: Brian Dean Message-ID: Date: Sun, 19 Jun 2005 05:15:14 GMT

Many (maybe even most) "robots" one comes across tend to be built with the primary purpose of obtaining a degree (MS, PhD, etc). Once they've served that purpose, development tends to stop.

I think that to make a great product, you have to stop thinking of it as a "robot", but instead think about what problem needs solved - what the potential customer wants. One owner of a drill company was purported to tell his sales force - "We don't sell drills, we sell holes." Stop getting all caught up in the bells and whistles of your product that no one really cares about and get inside the head of your customer and solve their problem. Then you have something that people must have.

Roomba sells not because it is a robot vacuum cleaner, but because it's a hands off vacuum cleaner that happens to be a robot. We have one and, contrary to some reports, it works great. No, it doesn't compare to our Dyson Animal, but turn it loose in the family room and it does an adequate job, all by itself, and finds the charger and turns itself off when it's done. It's effective in that we just bring out the Dyson less frequently for deep cleaning.

The really great "robots" may not be considered robots at all by many. Dishwashers and clothes washers. But the idea is the same - it's a machine that saves time and effort so you can do something else that you'd rather be doing. That's the killer app for automation.

But for purveyers of automation, it's really hard to compete with human labor as MLW's $400k security guard can attest. How about a robotic grocery shelf stocking robot - a potentially huge market to stock the shelves of stores around the world? But there's just no way a machine can compete with a high school kid working part time for a few dollars an hour.

The bane of automation is lack of environmental controls - which is pretty much the definition of the "real world."

So perhaps most "robots" are built to satisfy the creative outlet of tinkerers. Or to produce a degree. Or both.

Yep. And where it's not saturated, the killer app just hasn't yet been invented.

What we need are some practical, low cost super conductors. Something as easy to buy and use as a spool of regular copper wire. That and about a 10-fold improvement in battery/power technology. What a boon to invention / engineering that would be.

-Brian

You would think, but you know, the vast majority of startup companies go out of business. I have also been involved with companies that have been successful. I pride myself with the fact that I usually accomplish what I set out to do. Sometimes accomplishing the technology isn't enough.

In technology, hardware or software, you have a choice. Do something interesting at a startup, or do something boring at an established company. There are few exceptions to ths rule. An established company is more secure and generally pays OK with benefits. Startups are like playing the lottery.

To me, at least, a "nothing technology" is one where nothing is actually accomplished. A "roomba" is a small goal seeking robot that is a very well understood task. People have been doing it for 20 years. A carpet sweeper has been around for over a century. So what? they add an electric carpet sweeper to a micro-mouse. Snooze.

One out of?

Exactly.

If there is a "clear cut need" for a product, it seldom takes a visionary to think of it. Most of the time it is bolting two or more well understood technologies together. Say, a carpet sweeper and a micromouse.

20 years ago when they had micromouse contests, it was "visionary" and "revolutionary," but you couldn't sell a roomba for $1000.

Sorry, yes. When I say robot mean "autonomous mobile robot."

A never ending cycle, for sure.

You steal a robot. I thought that was implied.

I wouldn't say "all" are morons. I worked as a security guard when I was young. Even if not a moron, it is a job that requires only moron level intelligence.

A moron is not nessisarily a description of the person but of the minimum requirement of the person who will do a task. It doesn't matter if you have hired Tesla if you use him to push a broom.

Minimum wage hourly workers are pretty cheap. The functionality of the robot must exceed the capability of a moron and cost less than the minimum wage, otherwise business will hire minimum wage workers. I am dubious that the echnologies are there yet.

Aibo?

Brian wrote

The really great "robots" may not be considered robots at all by many. Dishwashers and clothes washers. But the idea is the same - it's a machine that saves time and effort so you can do something else that you'd rather be doing. That's the killer app for automation.

He makes a valid point. It also suggests some of the limitations of a robot. Fairly simple actions by "robots" have been successfully automated for years. But more complex actions still present challenges.

As an example of a simple action, I have an electric stove with a self- cleaning oven. When I push the right buttons the stove "knows" to lock the oven door, turn the power on the oven to bring it up to approx. 600 deg F, maintain that temperature for 3 hours, shut off the power, and unlock the door when the oven temperature drops below approx. 300 deg F. That's fairly simple programming in a controlled environment, i.e. the oven is literally insulated (and figuratively isolated) from the surroundings.

As an example of some complex actions, I recently purchased a Nikon D-70 camera. It's a great camera and I've already taken some good shots with it. To an extent, the camera is a robot, I can just set it in "auto" mode and snap away. Of course I didn't have to pay $1000 to just get a snap & shoot digital camera. Much of the value of the camera comes in the ability to customize the settings. Unfortunately, because there are so many variables, the programming is complex.

One of the complexities is that the camera must interact with the environment. I tried to take a picture of the sky at dusk, as the first stars were becoming visible. At first I had the camera set on "auto", the camera didn't take the shot. I just got a message that the light was too low. I then changed the settings and got a fairly good shot. But the point is, the programming correctly recognized that there wasn't enough light for most shots under those conditions. Until I specifically told it to make an exception it didn't "know" what I wanted.

I'm probably more knowledgeable about photography than the average digital camera purchaser, yet it will take several months to become proficient in the use of my digital camera. Consider a robot of the kind seen in sci fi movies from the 1950's and '60's. You could simply ask one of those robots to take a picture or make you a hamburger, and they'd do it.

Assuming that you could somehow design and build a robot capable of such tasks, how much programming would the consumer have to do to in order to get the robot to take a picture or make the burger the way he/she wants?

If my son makes a hamburger for me today (Fathers Day) he knows, or knows to ask: where the meat is; what size burger; seasonings; bun- and if so toasted; grilled or cooked on the stove; condiments; beer to drink ?. And if I tell him, he understands spoken English, I don't have to use a complex computer menu to communicate.

Give our current state of technology, robots may be helpful for well- defined, repetitive tasks. A Robbie-the-Robot type character is still far beyond our capability. If you really want to go into the robot business, think about some particular task and how automation could make it easier.

Richard

< snippy snip > < snippy snip>

I think the Balbot hit it pretty well. I think Jim Frye's modularization is a great thing.

One problem about the killer app, is that I am willing to bet that nearly every one of us has a "pipe dream killer app" tucked away at the back of our brains that we do not want to see capitolized on, lest we do not get a chance to do it for ourselves...

On that note, I will open up my killer app vault:

1) A true inertial balancing robot, with encoded motors (not wheels) 2) A back end replacement board for R/C servos ( see colin mackenzie ) that uses a more common bus and enables a wide variety of tuning and readback for position.

just my $2,000,000.00 worth.

Thanks, will take a look at it.

Wayne

.===snip---

Now I remember. The Turtle appeared way back when the TRS-80 and first IBM PC appeared. Then when Autocad appeared they used Lisp and maybe still do. I never got into either, so maybe now is the time to take a good look. Interesting that they are still around, even though not well publicized. Kind of like BASIC.

Wayne

(Repost of an article posted in comp.robotics.misc)

From: "blueeyedpop" Message-ID: Date: Sun, 19 Jun 2005 13:32:52 GMT

< snippy snip > < snippy snip>

I think the Balbot hit it pretty well. I think Jim Frye's modularization is a great thing.

One problem about the killer app, is that I am willing to bet that nearly every one of us has a "pipe dream killer app" tucked away at the back of our brains that we do not want to see capitolized on, lest we do not get a chance to do it for ourselves...

On that note, I will open up my killer app vault:

1) A true inertial balancing robot, with encoded motors (not wheels) 2) A back end replacement board for R/C servos ( see colin mackenzie ) that uses a more common bus and enables a wide variety of tuning and readback for position.

just my $2,000,000.00 worth.

How about a robot floor scrubber? I would buy one of those, and I have never seen one advertised.

Yup, it ain't easy.

It doesn't, however a pet rock may fund something better.

It is a debatable point. Also, may auto makers went out of business and were not absorbed.

Didn't it? All the capabilities were there. I would say the gateway technology was batteries.

That would be illegal and almost certainly cause someone to be convicted of man slaughter if not murder. If you knowingly create an intentionally lethal booby trap, that is murder.

Yes, but 4 morons can push a broom 6 hours each for $5+ dollars an hour with no benefits.

Mobile robotics is not at a stage where it is reliably turn-key.

Depends on the structure of the employment.

mlw wrote in news: snipped-for-privacy@comcast.com:

Have you ever owned your own company?

[snip]

So where does "nothing" turn into "something"?

All auto manufactureres currently in business are survivors of a "real" technology revolution. I will be so bold as to estimate that over 95% of the auto makers who no longer exist were absorbed by others. If you sell out of a business, I consider that a success.

[snip]

If you mean you couldn't sell a Roomba 20 years ago for $1000, you are right, it did not exist.

You steal a robot. I thought that was implied.

Easy enough to prevent, rig it up like the Talon units being used by the US military. If captured, self destruct, taking your "captor" along with you.

He couldn't puch a broom 24/7/365. This is what I believe is the key.

How much do you believe a minimum wage worker costs?

Aibo, target market, man in the street. (Actually, man in the streets spoiled child).

Yes they are, and that is sort of the point I was making. True revolutionary innovation is seldom successful

mlw wrote: [...]

Are the makers of the roomba robot successful?

Often "techie" people have no business sense or any understanding of human psychology or even the economics of their "visions".

When products like this come on the market they might create a better acceptance for the idea of autonomous appliances thus paving the way for the "something technology" that people such as yourself will bring onto the market?

They might even provide cheap hardware, due to mass production, on which the "something technologies" can take advantage of. Something like the white box, if mass produced, could be a basic unit for a "something technology" which otherwise would be too expensive.

The PC is a good example of a powerful product made cheap by mass production. I don't know if you see "MSWindows" a "nothing technology" but at least it created the need for mass produced cheap hardware to run it.

Apparently streamline cars were designed a long time ago but people weren't ready to accept anything that looked so strange. They were used to the horse and cart. Everything has to happen in stages and the "nothing technologies" might be a "something stage" in that process.

-- JC

Even worse than that, you would still lose a $400,000 investment if it blew itself up. And if it was self-destructed I doubt that insurance would cover it. :)

-- D. Jay Newman

Mark Haase wrote in news:mehaase- snipped-for-privacy@netnews.upenn.edu:

No, after 40 hours , you must, by law, pay overtime. Additionaly, though the minimum wage is $5.65 (or whatever it is this week), there are many more costs involved in employing people. FUTA, SUTA, insurance, advertising the postion, sick days, vacation, etc. Figure a labor burden of $10 / hour. That's a conservative estimate. 168 hours / week times $10 / hour = $1680.00 a week = $87,360 a year. Figure a 5 year lease, you could get a pretty good jump on your $400,000 robot.

Realistically, you will not find trustworthy people who will work for minimum wage.

When the price of a security bot gets to a level that a bean counter can justify, then it will happen.

mlw wrote in news: snipped-for-privacy@comcast.com:

I'm not too certain of that, since my pet rock died. I had him turned into gravel and spread him in the driveway. He would have wanted it that way. Alas poor Stony.

I'd need to see proof of that. Some of the very early automakers may have had their assets turned back into buggy making companies, but I would venture to guess that after about 1915 or so, with the dawn of specialized equipment not suited to any other industry, most automakers who went out of business sold their equipment to other auto makers. The market was starting to explode.

If the technology was there, and the perceived need was there, where was the Roomba?

Yeah,yeah. Details, details. All right, how about this? The robot "incapacitates" anybody who attempts to do it harm.(OK, we're messing wiht Asimovs incarnate laws here.) Wait a minute!! This is a "security robot", it's supposed to do something when there is a breach in security, right? Don't you suppose it would be programmed to take some action to protect not only the area it is guarding, but also to protect itself? Maybe it can dial 911.

Where do you find these people at these prices?

Though the minimum wage is $5.65 (or whatever it is this week), there are many more costs involved in employing people. FUTA, SUTA, insurance, advertising the postion, sick days, vacation, etc. Figure a labor burden of $10 / hour. That's a conservative estimate. 168 hours / week times $10 / hour = $1680.00 a week = $87,360 a year. Figure a 5 year lease, you could get a pretty good jump on your $400,000 robot.

Realistically, you will not find trustworthy people who will work for minimum wage. Soft costs, such as theft, are hard to work into the equation, but with the robot, they are eliminated.

When the price of a security bot gets to a level that a bean counter can justify, then it will happen.

Joe

If it wasn't clear, the proceeds from the sales of pet rocks, i.e. the roomba, if successful, may be able to fund more capable products.

There were, and are, many auto makers that you have never heard of. Small specialty companies. Look up factory five. This type of cottage indistry has been around for a long time. It's sort of like beer. With the exception of prohibition, there have always been huge numbers of companies, some profitable, some not.

What makes you think it would suddenly appear? Wht about wine coolers? People have been mixing b-grade wine with fruit juice for centuries. Suddenly in the '80s a couple guys think to sell it as a product. The a few product engineers decide to use clear malt beverage (zima). Thus the "cooler" was born.

That isn't the point. The question, can one or more security guard robots roam a facility in a more cost effective manner than a couple morons?

The robot will almost certainly require some sort of support contract or guy employed to make sure it keeps working. It will "never" work 24x7. Even if it worked 100% perfectly, what is the battery drain/charge cycle?

Everywhere.

If you hire more part time workers and don't pay benefit or insurance you can reduce he cost. If you hire cleaing company or security company, they'll hire illegals and shield you from the legal issues.

Lock the doors and and put cameras in place.

Yes, that is sort of my point. It isn't there yet, probably not for some time.