White Box Robotics Pre-order

I'm still debating the tracks for the stairs, and consdering many other solutions.

And given the state of Lee's health, a single-story house would have been a *really* good idea. Right now we have a townhouse and two hospital beds in the living room. This gives Groucho very little room to move about.

Actually I'm more concerned with the clutter than the stairs for a robot.

-- D. Jay Newman

No, price is a very important part of most market plans.

Exactly.

"my" robot is intended for hobbiests, and is nothing more than a "HOWTO." Whitebox is a "product" that really doesn't do anything.

Yes and no, a "modded" PC is typically a gaming PC, and high performance video cards that suck up power, lots of ram, and very fast CPUs, so I bet that it isn't even a very good modded PC.

Only as a "what if" scenario.

The tone of this paragraph is humorous, "tell them how wrong they are to use Windows." There are many ways to look at this statement. Think what you will.

Besides, I was interesed in the "robobusiness" thing until I realized it was about "robotics business" and not "robotics." It is the object of business not to do anything original.

You're right. Building a rectangular base with two gear motors from a ride-on toy, and strapping a motherboard on it, has never been done before.

-- Gordon

Ok. I will just have to wait and see.

Looking at your most recent pictures the motors appear to be scrapping the floor, almost?

Have you decided what kind of obstacle sensors you will be using and their arrangement?

I obtained similar toy motor/gears/wheels and have decided to build a light weight robot base as well, with a handle on top so it is easy to carry around. I will be using MSDOS until I know enough to port the code over to a Linux OS.

Again. I will just have to wait and see :)

- John

Buy the stuff, use a hacksaw to cut angle aluminum, drill some holes, install Linux, download and install the software, done. Yes that is a simplification, but basically it should be reduced to a fairly simple process.

No, I have a good 3 inches clearence.

I'm looking for some cheap ultrasonics, and thinking on how to do the bumper system.

I would go right to Linux, and if you want I can help you port the code. You tell me what DOS API you need and I can help you implement it in Linux.

Yes, mechanically, it is very simple, but that isn't the issue, at least not to me.

Where I think I am being, if not original, at least different, and I'm still thinking about the differentiation points, is the idea that it *is* just a PC.

I've spent a good amount of time (years) implementing and studying control systems and data acquisition, while not directly robotics, certainly applicable. There are a lot of things that are conventional wisdom about what must be done on real-time operating systems or micro-controllers. The fact is that many of these applications can be done easily on a PC running a reliable OS.

It seams that I have hit more than a few nerves with this attitude on this group, as some of the bigger flame wars have indicated.

Take PID for instance, it is conventional wisdom that you can't implement it well on a PC running a standard operating system. Everyone says I'm going to have to put it on a micro or use a real-time OS, but really, there is no reason too, and a small number of calculations 50~100 times a second is hardly any sort of load on a modern PC. This information is not out there.

In my career, I have been sort of a problem solver type, and there have been few problems I could not work around. I remember at Metrabyte, I was able to get a DAS1800 to stream full speed to a disk file under Windows NT 3.51. At the time, the 233MHZ pentium was fairly standard and the DAS1800 was an ISA bus card. It had never been done before, and previously the only way to do it was to use a realtime OS (I think QNX was out at the time, I don't remember.)

The point was that to accomplish the objective, you couldn't just focus on one aspect of the problem, you needed to take a more systemic approach. By studying the behavior of the OS, and learning its ins and outs, you can usually implement what you want.

So, what I think is innovative about my robot is that I'm approaching it from an OS and computer science perspective, and developing it as an application running on a PC.

This is very true, and you haven't demonstrated it yet. Not by a long shot. You've developed one or two pieces of isolated code, and have yet to implement any kind of sensor system, especially using a hybrid of sensors as is the norm today. You have yet to implement speech and recognition, vision, and other control dynamics, running them all concurrently. What you have so far is no more a "system" than DOS running VisiCalc.

You also continue to mistate what other people have said about the use of microcontrollers. The issue was scaling the system and still being able to handle the wide variety of common robotic functions. No doubt that when you encounter load limitations, you'll back off on features, and claim the design is done. If anyone then tries to add anything, and latency problems occur, it'll be their programming that will be at fault.

-- Gordon

I've just started this project in earnest a few months ago, and I think I've made some very great progress. Using largely off the shelf technology, I have a system that moves by remote control, has a video camera, a dual wheel opposed drive that can maintain a straight line based a PID system that works very well with variable time slices.

In good time.

I actually have it speaking. I am looking at something like IBM's ViaVoice for recognition.

By "system" I was refering to was the system as a whole, the operating system, the motherboard, and the application.

Please site an example.

That was one of the issues, yes and as I said a few calculations 50~100 times a second are no load on a PC.

No doubt? There is of course a finite amount capacity in any system, so obviously if I never stop adding tasks for the robot, then I will eventually run out of capacity. Will a hobby robot ever be finished? Every robot will eventually be out strapped by additions, so it is inevitable that I will eventually run out of capacity. That doesn't mean you are right. My system will, however be very expandable.

You accuse me of misrepresenting what people have posted, when you are probably one of the worst culprits. I have never said micro-controllers do not have a place, never once. I have said that I have no interest in developing micro-controller code and have no desire to use micro-controllers in a capacity where their function is not well defined. My phrasing may have been different in different posts, but the intent has been consistent. You have taken this to mean that I don't want to use any micro-controller based technology, which isn't true. A self-contained stand-alone device whose behavior is predictable, well defined, and cost effective would be acceptable (linksys router, hard disk, etc.) Again, I've said as much in other posts.

Are you saying that there is *any* system on which no program can screw it up? Of course not, there are always issues. You're probably a Windows user, would you ever stream video to disk, burn a DVD, created an index on a huge database, transcode video to mpeg, play mp3s, surf the web, and run a mandelbrot? Probably not.

The robot may be running on a general purpose computing platform, but it is not a general purpose computing application. There are always constraints.

I think you apply what is feasible for you to everyone. Sure, for some, this idea is great, for many others it is not. Its all simple to you, cause you have a certain background, the time and the drive to do so. To you and most of us on this group it sounds easy. Buy some metal, cut and drill it, find a couple motors, hack a mouse, etc. But for an ever growing # of people becoming interested in robotics, its not.

So your instructions say, go find a motor out of a toy car. This translates into "spend hours looking for one of these toys and hope you get one identical to mine.". What if it doesn't mount the same as yours? Now they are on their own to adapt your instructions to the motor they got. What if they didn't get the right voltage? What if their motors don't have same speed? What if the motor controller you tell them to get can't handle the motors? Oh, now they have to dispose of the rest of the toy. These are all simple problems for you and I to address. To start with, we know what to look for and we know how we could work around any differences. But for many, there is a lot of time needed to figure out and learn all this stuff that likely, they are not even interested in, especially when first getting into robotics.

Don't get me wrong, this is exactly the route some people *want* to take. But your opinion based on this thread seems to be that the Whitebox robot is a bad idea because everyone would rather spend a few hundred less and countless hours to have a similar (and inferior) robot.

As for the cost, yes, $1200 seems like a lot. But I don't think it is that much for what it is. The amount of time saved alone could easily make this robot perfect and cost effective for many. What about warranty? How about durability? How about tech support? Even the look is polished. What about quality? I mean, the motors, wheels, frame, etc used in the Whitebox are superior to your home brew bot.

Do you think your $500 is realistic for most? You are counting every piece of wire, connector, screw, stand off, cotter pin, washer, etc, etc? In addition to parts, many people are going to need to invest in tools. A lot of people don't have anything more than some cheap screwdriver let alone a drill, hacksaw, dremel, etc. Then factor in the time involved and the robot you end up with at the end and the $1200 is not looking so bad for many.

Will I be buying a Whitebox bot? No. But then I have the know how, the tools, the space, the motiviation and the *want* to build my own.

-C

Ah, yes, tools! That's a whole new can of worms. But for me, I look for anything that's an excuse to buy new tools.

BTW, nice to see the info on Lazlo again. I had lost track of it when you changed domains, and didn't catch where it all went until now. (Okay, I'm slow...)

-- Gordon

attempts to

Litte more than those? Have you had a chance to work with each one of the HERO robots and the other robots from the 80's? The HERO's were and still are capable robots which still have new options being developed as add-ons. There was also quite a bit of software written for each one. From the research i've done the HERO 1 alone sold more units than most of the other personal robots combined and helped spark interest in robotics and engineering to a lot of people (myself included). The HERO 2000 was a complete redesign from the ground up and is an excellent robot! Easy to program, fun to use, and looks great.

The RB5X is going under upgrades/improvements and is also a nice platform which could acommodate newer brains, etc.

Some of the other personal robots like the Gemini and Newton were also sophisticated and had a lot more depth to them than mere toys. A lot can be learned from them and the history of the personal robot companies.

While some early robots were glorified "toys" quite a few were not.

How can you comment on how useful something is when you haven't had a chance to see what it actually will do??

building.

Perhaps standardize all the algorithms and logic... With all the different platforms out there a generic version may be watered down too much to be useful. Getting navigation down is important.

Afterall, if a robot is mobile then it ought to be able to get around!

The "people like us" may just enjoy other aspects of this great hobby. I know several who would rather buy a viable platform then expand upon that. I like to do both!

For a finished product like the new WhiteBox it appears that it is marketed toward both robotic hobbyists and the masses. I've talked to Tom on several occations and he is very dedicated to robotics and is a nice guy as well. He believes in his product and is doing his best to see it take off. I've been involved in robotics for years and I wouldn't underestimate the importance of looks or having a finished product. That seems to be a requirement for anyone else than the hardcore robot hobbyists. Besides, it might even pull in and expose more people to the hobby that otherwise would have overlooked it...

control is,

I've enjoyed reading your posts and it looks like you have done an excellent job on using mouse encoders as encoders for your main drive. I look forward to seeing the results of that and what comes out of the rest of the project.

The one thing I can do without is some of the harsh comments which have popped up in these threads and the point of view that there is only one way. As I see it there are many excellent solutions to the problems in robotics and it is best to keep an open mind. Linux, Windows, microcontrollers, all have their place.

Just my 2 cents....

Robert

To take this one step further...

I'd be happy with a generalized interface to things like path planning algorithms, but I don't see an agreement on the best alrorithm to use in order to standardize on one. There are so many -- A*, D*, Floyd-Warshall, Voronoi, and bunch others. Some robot nagivation systems use several algoritms at once, to interpolate the data for comparison, learning which is better in different circumstances.

Some (many?) of these algorithms are canned public domain routines. You can download them from AI, gamer programmer, and other sites. There's no consistent interface, though. At the very least, as a DLL you could just give it data, and it returns a path as a set of numbers. Not everyone is up to the programming level to create their own DLLs. I can well imagine the popularity of these kinds of libraries for the large number of students who rely on Visual Basic, for example.

Given the disagreements over engineering practices vis-a-vis the use of microcontrollers to offload critical hardware functions, I can hardly see the value of shoehorning more one-size- fits-all methodology into the mix. Who's software? Who's navigation algorithm?

-- Gordon

Yes, I played with a hero when they came out.

Did it spark an interest or ride a wave? I worked at Denning Mobile Robotics in the early/mid 1980s. There was tons of interest in robotics at the time, even investors! I even built my first "robot" in 1979 out of a mattel bigtrack and an RCA1802. I don't think hero sparked an interest, it was just the best of a fairly weak lot.

Yes, it did look great.

I'd like to see that.

As a person who was involved in robotics in the early days, there is much that can be learned by seeing what we did 15~20 years ago.

Depends on how you define "toy."

Why is it that people assume a contrary opinion is uninformed? If you want to discuss robotics technology or robotics history, that's fine, but if you want to state your assumptions of what another person knows or does not know, then you are acting foolisly.

The problem with non-existing standards is that there is no standard reference to know exactly where the standard is lacking.

For instance, would an API call for robotic movement resemble a LineDDA or a bezier curve? Should robotic motion be controlled by vectors?

It would be nice to standardize on this sort of API, that way, my $500 robot will allow me to develop software that will work on a very nice $5000 robot without changing.

A nice API certification would be nice, where XYZ robots could comply with RMAPI 1.0 (Robotic Motion Applications Programming Interface). If a number of robot OEMs supported a standard, it would be quite nice.

ok.

And this is the probem. It is (IMHO) to expensive for the hobbyist and lacks any well defined task for the masses.

I have no doubt.

He better, he put a lot into it.

Defined finished? Shouldn't it also perform a function? Roomba actually sweeps the carpets.

I would say roomba does a better job exposing robotics to the masses. Too bad it is a fairly weak application.

Thanks.

I do find it funny, if you go back and read what I've posted you'll notice that while I do enjoy controversy and debating a contrarian opinion, I try not to be the who creates a flame war first. It is impossible to defend myself against this without reopening the controversy that you reference.

Ahh, yes, "every one has its place." I think that is political correctness. The debate is where that place is, and there are some serious issues involved.

What an incredible comment! Of all the '80s-era robots (discounting the fancy toys), the HERO undeniably sold the best -- a minimum of 20,000 units for the various versions, since the initial kit was first introduced in 1983. Entire robotics courses were created at the high school and university level around this robot.

Denning was incorporated just a year earlier, so it's hard to claim that HERO only rode the wave. The HERO was first to commercialize a number of concepts that the other players only announced in prototype form, including the trainer arm and voice.

You've already expressed your contrary opinion, so facts don't mean anything now, but there you have it.

-- Gordon

I never said it wasn't popular, I just doubted that it "created" any sort of wave.

Denning, as you point out was incorporated a year earlier. So, the "wave" was before Hero, wasn't it? Right?

So, if the wave that helped Denning, and a slew of other robotics companies, get funding arrived before Hero, it is only reasonable to say that Hero rode the wave as opposed to created it. Wouldn't you agree?

When Hero came out it was a toy, that was our perspective. We were working on a 480 pound mobile robot security guard with a 24 ultrasonic sensor ring, 6 floor based ultrasonics, phased array ultrasonics, video camera, a digital radio link, syncro drive, and radar.

Maybe you guys thought it was a capable robot, but many of us in the industry at the time thought it was little more than cute. It is a matter of perspective.

On an educational level only.

When you want to argue facts instead of making ad hominem attacks, I would welcome the change.

Enjoying a vigorous debate, especially a controversial or contrarian one, does not imply that I am not interested in facts. To make the charge that "facts don't mean anything now," is basically an underhanded way of you trying to impuge what I write, implying that I have been less than factually accurate, a charge you should appologize for making because you know better.

I may hold different opinions, but I know my facts.

Oh, right. Care to remember when Denning's Sentry product was announced? It was some six months *after* the HERO. Denning was still showing prototypes when Heathkit was already actively selling to schools and individuals. Besides the fact that Heathkit had been working on the kit for more than a year, and they essentially created the educational robotics market.

Denning had trouble selling their technology, and they went horribly in debt, but I have no trouble giving them their due for their contributions. But that's not the argument. The argument is your belief the biggest selling educational robot somehow never "sparked an interest." That has got to be the silliest thing I've read from you.

And then there's this doozy:

If "popular" doesn't mean "sparked an interest," what does? Who was buying $40 million worth of HERO products anyway?? No one with any interest in robotics, I'm sure.

-- Gordon

Heathkit was making an educational product, Denning was attempting to create a securty guard robot that lived up to an expectation of performing a task.

Yup, that is very true. They had terrible internal problems as well.

Yes they did, I had left the company by then. It was my first professional job in technology. I had my picture in the paper, a robot I built for testing published in a japanese robotics magazine, I met some interestng people. It was a blast. Like the recent DotCom bust, the robot bust came just as certainly.

I'm glad you said that. There was lots of stuff that went on at Denning. Most startups fail, mostly not because of the technology, and as you said we are not discussing it.

Oh please, people were interested in robotics before Hero, Hero rode the wave of the trend and captalized on it.

I think you are confusing "cause and effect." There is no evidence that Hero caused anything. The Apple iPod is widely popular but it had nothing to do with the digital music revolution, it simply managed to captalize on it. Heathkit was popular with educaton and hobiests at the time. It is diffcult to remember how big Heathkit used to be before Zenith bought them.

IBM did not spark the PC market, Apple, Commador, and kaypro did, (maybe even imsai) but IBM sold the most.

There are so many examples of where the products that sold boatloads had very lttle to actually do with the market place. Sales is not a reliable measure. You are looking back over 20 years. The real "spark" was probably R2D2 in 1977. When the pioneers are asked what inspred them, it is usually R2D2, the lost in space robot, or the robots from "silent running."

Nah. The real "spark" was Tobor the Great. Or maybe even the robotic Maria in Metropolis. Cute *and* dangerous. Yeah, that's it. We owe modern computing to the spark of interest created by the abacus. Oh, that and the female-sounding computer from Star Trek. "Comp-u-ting!"

-- Gordon