My company has funding to expand our product line into some new
areas such as robotics. We are willing to take some risks in
order to enter a new market. Our current expertise is electronics,
especially low-cost microcontrollers.
I would like to open up a discussion about designing the minimum-
cost robot that at least a few people will want to buy. The main
object is to break into the market and learn the needs of the
typical purchaser. Any and all ideas about features or about
other companies current offers (especially areas where we can do
better than they did) would be most welcome.
If this becomes a real product, I will let anyone who participated
in the discussion have one or two at cost, and I might even give
a few away.
Because this will go into production, suggestions to include
used or surplus parts won't be much help. The parts have to
be things that we can make or buy.
(I am crossposting this to misc.business.product-dev and to
comp.robotics.misc. Crosspost to misc.business.product-dev
from comp.robotics.misc are welcomed, so there is no need to
trim the newsgroups line)
Let me ask some business questions.
Who is your target customer? Are they hobbyists who'll buy a couple of
units, and maybe a variety of optional accessories? Are they companies
that will buy larger orders of integrated equipment? Are they academic
institutions who'll make students buy a kit every semester for a few
years? I ask because it makes a difference about how to design the product.
The hobbyists probably want a basic platform that can be customized a
bunch of ways - they'll want accessories, and a CPU that's amenable to
hacking at home (PIC or 8051 or AVR, in a DIP). Absolute lowest cost
isn't important, as long as the price seems 'fair'.
The corporate folks may beat you up on costs, and may rather have higher
performance parts (industrial temp ranges, low EMI, good connectors,
The schools want a low cost and a bunch of pack-ins for their lab
exercises, and the gear need not stand up to repeated abuse if the
students have to buy it. Cheaper connectors might make sense.
Are there already products on the market that target the audience you're
interested in? Maybe you can address the shortcomings of the
competition (price, performance, features, etc). Or is this a new
product going for a new market segment?
I'm not a big robotics hobbyist, but I think the Lego MindStorms are
pretty cool. Not cheap, but ya gotta love the dev tools.
That's an excellent question. I am inclined to target hobbyists with
the first version, and then decide what direction to go. This is a
"test the waters" idea at this point, so selling just a few is OK.
There are also some hobbyists who aren't ready for assembler, so maybe
a BasicX-24 would be appropriate. (See
Another alternative would be to provide a radio link and let them
write programs on their PC.
I would love to find a new market segment, but that's easier said
Rafael Deliano wrote:
Directly. Order by web page, deliver by mail.
That's a really interesting design. Something like that would do
Oddly enough, I worked with the creators of that zeppelin on
a related project.
I have spent too much time making toys for Mattel already. Now
I just want to make and sell a few interesting low-cost robots.
That's a *really* good idea. I wonder if i could pull it off?
The electronics/optics I could do in my sleep, but it would
take some clever mechanics to make a low-cost micro-manipulator.
To get back to basics its often good to
step back in time. The book
Eder "Moderne Messmethoden der Physik Band 1"
VEB Berlin 1960
has already about 3 pages on the subject
and claims they have been around for 100 years.
Some are purely mechanical ( pantograph ),
but most by that time use electric motors or are
One is based on thermal expansion of wires, but
Nintinol was invented only later in the 60ies.
Big Manipulators/Teleoperators got off in the
60ies in the nuclear industry as is well known.
Köhler "Typenbuch der Manipulatoren / Manipulator
Type book" Thiemig 1981
lists available units but is thin on the
McCloy, Harris "Robotics - An Introduction"
Open University Press 1986
has a chapter on it, but most modern literature
on robots avoids the subject.
The shop where the Asuro "educational toy"
originates from DLR has a "Robot"-program ROTEX
that is actually manipulators for use in space.
Wayne Lundberg here, co-founder of chapter 299 Robotics of the Society of
Manufacturing Engineers back in 1979 when robotics became an interesting
concept for replacing union workers and their eternal strife with devices
that did not demand impossible stuff and could work in a lights-out facility
24 hours a day, 7 days a week.
County area to companies such as Rohr Industries, Solar Turbines (a division
of Caterpillar), Sundstrand, General Dynamics, and a whole bunch to Cypher
Data. An analysis conducted by students at San Diego State University
concluded that of the 27 high-end robots, only one was able to even come
close to a return on investment; the one a General Dynamics due to the
never-ending efforts of Mark Raptis, a devout believer in the future of
robotics. The bunch at Cypher Data nearly broke the company and the
cartesian coordinate IBMs were given to SDSU's engineering department. Where
they still collect dust and nothing else.
Robotic applications are strong in the electronics assembly field where
surface mount technology is now dominant. Another broad application for
robots is in auto manufacturing, in spot welding, regular welding and in
As to hobby robotics, I have a 7 axes Rhino robot with controller, offline
programming, PLC built in, and all the works which cost $8,000 back in 1987
and for which I have yet to receive a single offer from many postings in
various newsgroups. I have yet to put it up on eBay because when I research
robot sales on eBay I see a disappointingly few and worse, the money the
prospective buyer is willing to pay is not worth the box I'd have to package
the robot in.
I propose that you create a number of keywords, put them in eBays closed
auction search functions and see what people are willing to pay for robotic
devices. This is the latest and greatest marketing tool never available to
anybody until just a few years ago. It is so revolutionary that no marketing
profession will even talk about it because as soon as their clients find
out. they, the marketers, are out of business since anybody can do what they
charge thousands to do.
I hope this helps you zero in on your target market. Feel free to consult
with me at your pleasure. No fee, no charge, just in the spirit of sharing.
Another thing. At Solar Turbines, division of Caterpillar, we got the City
College to develop an automation course heavy on PLCs and Robotics. We then
tried to advertise the course at various manufacturers in the area. We were
forbidden to do so because the prevailing unions would not open the door to
a classification with the word automation in it. (Ever wonder why almost all
manufacturing is going offshore?).
all of the following is *opinion*
There are two types of robots,
small toys and medium to large
Are you going to go for a small AA
battery powered, microcontroller
based rover? RC car chassis
with a micro? I find these of little
practical use except to program, and
being so small, they usually lack the
type of sensors I want to program.
I think it would be best to have a
medium sized platform, with
a large area for 12V ?motorcycle?
battery. Balanced with battery
weight very low to the ground.
2 driven wheels, caster on
front and back. Caster wheels
need to be large to get over
bumps and through thick
carpet. Driven wheels are
centered on robot, so when
spinning the front and rear
of robot are equidistant from
center of spin.
Encoders on wheels/drive
At least one sonar sensor
on a servo. I think the
cheapest way to get this done
would be to get SRF-04 and
program a PIC to talk to it.
Have PIC talk I2c to the micro.
Include (in PIC) code to turn
turn the servo with PWM.
cheap IR detectors (3 on front,
3 on rear, 2 on each side)
so it doesn't have to bump
a wall to know it's there.
microswitch bumpers at
each IR detectors, in
case they fail to detect
^ ^ ^
| | |
| | |
V V V
In my opinion, there are already
enough line followers/programmable
turtles out there. If you design a
new robot, it needs to be autonomous.
PID? on motors. Can this be done
with a PIC?
Keep everything hackable and
expandable. Perhaps offer an
"upgrade" that is brackets/
plastic casing, to make the robot
BASIC would be great for people
to learn how to program...
Micro should have some RAM
and storage device.
option of wireless link to PC, so
you can do remote control,
view sensor/motor status,
program the micro from the PC.
cheap low resolution camera
for vision. Line detection, object
recognition, docking with charger.
software! If working with a
wireless link to the PC, use
MS or other speech recogn.
program to allow voice control
of the robot. Autonomous
mapping without needing
A website where you can
share code. Download
a ZIP file, unzip and send
to the robot. Include
sensor updates, mapping
program updates, etc.
Guy, as you will see from my earlier post regarding possible markets for
industrial robots... I am a likely customer.
I have, for years, searched for a cheap and easy interface between a
standard PC and a PLC like device to drive either servos or steppers plus
sequential programming of switches with time elements, etc. I have a Minarik
PLC, a model 100 Radio Shack at one time the leader in this field, and half
a dozen PCs for normal work. But nothing that I could connect an RS232 or
Centronics or the new interfaces to a kit I could program using simple
language such as what once existed in Lotus: "Factory". I even have a BASIC
Stamp, but found it just too complicated for my short-circuited brain. It is
fine for the young and wise among us, but for atrophied brains like mine....
Leggo has a beauty, but it's limited to very simple commands... from what I
I'd be your first customer if you could come up with something like:
Make your PC talk to all the lights and appliances in your house. Make it be
your burglar alarm and secret spy. Turn on water and turn it off. Connect it
to infrared controllers in your living room to start your lawn-mower and
turn off the coffee pot. Use it in your shop or factory to drive conveyors,
monitor your heat treating oven......
All of this for $99 and ready to run by simply connecting the fully
assembled kit, including two servo motors (or steppers) with all wiring,
instructions and warranty.
I'll send you Paypal when you come up with something like this.
The other problem with novel is that people will not recognize it's value
and like any paradigm, will not see it at all. This makes it tough to
market. Look at my son's great idea and wonder why it has not captured a
million buyers to date. It's just not in a person's mind.... a remote
controlled sailing model for land... not sailboat, but sail car.
and note that everybody that does take a look at his
model fall instantly in love with it. But ten seconds later it fades from
their mind because they can't associate it with anything they are really
familiar with. So it will take a celebrity to be shown with one of his
models somewhere before the rest of the world can 'see' a radio controlled
surface sailing model.
Alas, I fear that you are right. We are exploring a number of
possible new products, and emotionally I would like to make a
robot simply because it would be so much fun to do. That's a
perfectly fine basis for starting a newsgroup discussion, but
not for spending real money developing something.
On the other hand, my management is OK with making and selling
something that will only make a small profit or even a small
loss, just to test the waters. I am still hoping that someone
on comp.robotics.misc has an idea that will lead to a product.
Wayne Lundberg wrote:
I use eBay a lot, yet it never occurred to me to do what you describe.
I just did a couple of test searches, and got some quite useful info.
Guy Macon wrote in
Design something like this on a smaller scale:
Either in a hobby kit form, or as a personal mobility device for the
disabled, to take the place of a wheelchair.
If you've ever seen someone in a wheelchair stuck outside a building
with no ramp, or stuck in snow or mud, or having to wait at the car
while others go hiking, I think you will see a market, albeit a small
one. Insurance companies pay for most of the cost of wheelchairs if
the doctor "prescribes" it.
I would buy one.
Hmmmm. I believe you need to rethink thisapproach. I feel the best, most
successful products are those that the inventor/builder/marketer feels
strongest about, and those tend to be ones created from their own idea.
You've been in robotics long enough to know what you personally like and
don't like. This is the best gauge you can have as to the type of
product you should develop. Create a product YOU want to see. Hopefully
it'll be what others want, too.
This doesn't mean you can't ask for customer input, set up focus groups,
etc. But the product needs to be one that you're willing to "bet the
farm" on. Granted, this approach puts more of you on the line. You're
not as able to say, "Well I guess our customers steered us wrong."
Mike is correct about the oversaturation of the market, particularly on
the hobby end.
[snip about robot markets]
I have been battling with this very question. I was involved with a fairly
spectacular robotic company failure, "Denning Mobile Robotics." I have also
been in several companies that have gone belly up since then.
In all the companies where I have had a direct influence one attaining our
technology goals, we have. The disheartening aspect of this is that the
success or failure is hardly ever to do with the technology. The iRobotics
"Roomba" is a prime example, it is a nothing technology, a carpet sweeper
monted on a goal seeking mouse robot.
Companies with "real" revolutionary technology seldom succeed. For every
example of success, at least 10 or 100 failures can be found. This is
because the technological visionaries may come up with great ideas, but
they are usually abstract, and focused on their field and not the general
public. It is not until the concepts filter out into fields not directly
related that they get applied to solutions the original innovator hadn't
suspected possible (or had the knowledge and/or ability to accomplish)
Robotics is just as hard. We can do some "interesting" things, but actuall
non-trivial applications that really "work" are a long way off.
At Denning, we all loved robotics. We were all passionate. What we didn't
know we learned as quickly as possible. We built some cool robots. We built
a security guard robot. It could roam a factory or stock room, look for
intruders, and notify the proper people of irregularities. It had
ultra-sonics, radar, video, and a data link. It was harder 20 years ago
than it is now, but the same capabilities exist. Suffice to say, a few of
us on light funding embarrased MIT and Carnege Mellon, and produced a robot
that could do, in real time, what their PHDs projects had to do in stop
We thought the "security guard" as a great application. It presented few
problems. We thought we could deploy it. We thought we could sell it. We
thought we were going to be rich. Until .... Someone asked how much it
cost. We said $400K. They said that's rediculous, they could hire a lot of
minimum wage workers to do that job. If a minimum wage security guard is
killed it doesn't cost the company a capital asset. Also, the robot would
probably be the most valuable and most mobile object in the room. In short,
it could do the job, but not in a more cost effective way as a moron.
So, mobile robotics is a technology that is growing, but it is still not
very cost effective.
if you want to sell a robot, you are probably aiming towards schools or
hobbiests, in which case you need to make it flexable and cheap. You need
to focus on the development environment. Lego MindStorms already has the
small robot market, so you'll probably want to focus on one that can carry
That's just my opinion, of course.
mlw wrote in news:t email@example.com:
I've never heard the term "nothing technology" before. Can you elaborate?
Is there a cutoff between a "nothing technology" and a "something
I assume you mean "10 to 100".
This is true of the business world in general.
Not neccesarily. If there is a clear cut need for a product, and that
product can be produced cheaper than it can be sold, it will be developed.
Or else the government will develop it.
If you mean autonomous robotics, I agree.
Geez, I'm still learning.
We built some cool
No, it just drives their insurance premiums way up and causes lawsuits. How
do you kill a robot?
Therefore, all security guards are morons. Hmmmm...
It is up to the collective "we" to bring that cost down.
Aibo. Target market?
And mine also, of course.
One that has a wireless connection to a PC to take advantage
of the computing power of a PC and the ease with which it
can be programmed using BASIC or C to get the robot to do
"interesting" things without the user having to be an expert
on the Windows OS or the Linux OS.
Regardless of the on board computer used a useful robot
should have a wireless connection to the home computer
to make use of its power and ease of programming to get
the robot to do "interesting" things.
A bonus for a non-professional programmer would be the
availability of library to make use of any on board
sensors or actuators for programming those devices and
a tutorial showing how to use those libraries from
Java or even C or BASIC.
mlw wrote (snip)
Mlw, why did you post to comp.robotics.misc and misc.business.product-dev
with followups set to comp.robotics.misc only? I assure you that posts
about this topic are more than welcome in misc.business.product-dev.