Results 1 to 10 of 10

Thread: I, Gömböc

  1. #1

    Post I, Gömböc

    (continuing from here - my thanks to jes1510 and to Adrenalynn who suggested I start this thread.)

    I want to build a robot that can come with me when I go to work. I write software for a living, so I always carry my laptop around wherever I go. My laptop is small and light, and so is its case. Still, being the lazy person that I am, after many years of carrying it around, it got me thinking. "Wouldn't it be great", I thought, "if I had a robot to carry my laptop for me?" And then I thought "Wait a minute! If I had a robot, why would I need a laptop? A robot is, after all, a computer!"

    That's what I want to build. I want to build a robot that can follow me around, carry my stuff, and be my computer whenever I need it.

    Requirements
    • It has to be as light as can be. Weight costs resources that can be used to prolong the robot's uptime. Also, I want to be able to pick it up if I'm in a hurry.
    • It has to be resilient. The real world can be rough so it has to be able to withstand a few bumps and bruises.
    • It has to be strong enough to carry a couple of kilos of stuff placed in its compartment.
    • It has to be water resistant. I live in the Netherlands. It rains a lot here.
    • It has to be able to follow me around, and not confuse me with other people.
    • It has to stay operational for at least 4 hours between a charge or a battery replacement.
    • It has to be able to climb stairs. Because I do.
    • It has to be able to communicate with me. It has to follow my lead, and to tell me when it cannot do so.
    • It has to be able to connect to my wireless network.
    • It has to be able to connect to a mobile network.
    • It has to be able to find its own way when I'm not there. I may want it to do my errants.


    Design
    I haven't decided yet, but I think it should look like a "Geochelone elegans", also known as a Star Tortoise. The star tortoise is very close in shape to a Gömböc, the first known mono-monostatic object - that is an object which is always self-righting. That way, if it ever looses its balance, it should fold its feet under its shell and wait until it rights itself and it feels stable again.


    So how do I go about building such a robot? Well, I don't know - yet. But here are a few first thoughts.
    It'll need a lot of processing power, so I'm considering an Atom motherboard, perhaps the IM-945GSE by MSI. I like the Dynamixels a lot - but I wish they were smaller.
    I like Li-Po's for power, but I have no idea how I should go about connecting them to a motherboard.

    If you would like to help, or for any other reason, feel free to contact me. I do need lots of help - this is going to be my first robot, and it's a big project. I, for one, promise to publish everything I make. (More on that later.)

  2. #2
    Join Date
    May 2008
    Posts
    2,228
    Images
    155
    Rep Power
    125

    Re: I, Gömböc

    I hate to be a pessimist here... but this is the sort of thing that entire University teams have tried to do, and failed. Of course I've pretty well earned the title "Mr. Reality Check" around here... so take this with a grain of salt.

    Lightweight, able to handle a beating probably isn't terribly difficult to do.. it's gonna be tough to design something that can go long distances, and up stairs, and still have 4+hours of battery life on board. Long distance of travel usually mean wheels... but wheels+stairs are typically bad. Building a walking robot that can carry 4+ hours of power probably means you have to go the route of hydraulics, or look at a gas motor... batteries would probably cost the same as a small car.

    Follow you around, probably best done by carrying a transmitter of some sort on yourself + visual tracking. Still, in a crowded area, might be tough. Generally speaking, in robotics, it isn't too hard to make something that works 25% of the time, but designing a system that works 100% of the time is near impossible....

    Find it's way there on it's own, every time... that's an entire PHD right there.

    -Fergs

  3. #3
    Join Date
    Apr 2008
    Location
    Sacramento, CA, USA Area
    Posts
    5,341
    Rep Power
    174

    Re: I, Gömböc

    What's your budget for this project? Barring some of the points that Fergs has so aptly pointed out, I think we need to get a feel for what you're looking to spend. At first blush I'm thinking $35,000 USD should be your starting point. That will get you the base and batteries pretty well squared away. From there, sky's the limit.
    I Void Warranties�

  4. #4
    Join Date
    Sep 2008
    Location
    Albany NY area
    Posts
    22
    Rep Power
    37

    Re: I, Gömböc

    I'll have to agree with the others here, this sounds very ambitious. More ambitious than stuff I've seen proposed to funding agencies for millions a year in research dollars. I'd start out with something a bit simpler, perhaps building a robot to one or two of your specs and seeing where you can go from there. In fact, building a series of robots each of which emphasizes a few of your desires will keep you occupied for many years, and you'd be learning valuable lessons in engineering which can carry forward to more ambitious projects in the future.

    A few very hard tasks you mention:

    "It has to be able to follow me around, and not confuse me with other people." would probably require some kind of special tag on you and a receiver on the robot - current vision systems aren't good enough to not get confused by other people or even objects in your vicinity. Consider a radioactive tag if short range, (lead suit costs extra), otherwise you'll need something highly directional that doesn't tend to multipath (bounce off objects and create ghosts), which means radio is pretty much out unless you have a lot of robots that can triangulate. Or you can wear a suit with some kind of unique pattern, but good luck being seen in the rain, from a distance, etc.

    "It has to be able to find its own way when I'm not there. I may want it to do my errants." is also well beyond the state of the art. Consider that GPS systems don't have the resolution to get accurate enough for, say, step placement. So outdoor localization will be a large problem for you - you may have to invest a bit in the infrastructure setting up registered beacons. As for "doing your errants (sic)", look at three-layer robotic architecture that includes an explicit symbolic mission layer. I suspect you'll need to invest a few dozen man-years to get much beyond simple tasks (don't expect it can go to the grocery store and do your shopping, but it might be able to go to a fixed location and pick up a package you've prearranged to be placed there and have it transported without much work).

    "It has to be able to communicate with me. It has to follow my lead, and to tell me when it cannot do so." This one can be easy, if you don't go for NLP and stick with some kind of wireless link. Following your lead we've already mentioned; the simple solution would be a remote with a tether (not wireless) - strain gauges in the tether can indicate where you are relative to the robot. That would also help with the power problem - you can wear the 100lbs of batteries instead of the robot!!
    --
    Gorbag
    We are pattern matching machines

  5. #5
    Join Date
    Apr 2008
    Location
    Sacramento, CA, USA Area
    Posts
    5,341
    Rep Power
    174

    Re: I, Gömböc

    I just noticed another design criteria. You want a tortoise that can climb stairs. I've had various turtles in my life and the only way they navigated stairs was in the jaws of my Jack Russel TurtleTerminator... I'm wondering about the purpose of that requirement?

    >> I like the Dynamixels a lot - but I wish they were smaller.

    The only way you're carrying that kind of weight around, especially on stairs, is with WAY bigger servos.

    I wish I were a supermodel. Alas, there comes a point when we realize that we have to accept basic universal truths. One relates to microservos carrying multiple pounds of payload, and the other relates to me being a supermodel. And neither are pretty - literally and figuratively.
    Last edited by Adrenalynn; 04-25-2009 at 07:23 PM.
    I Void Warranties�

  6. #6

    Re: I, Gömböc

    @lnxfergy:
    As I said before, my ignorance knows no limits, and I know I have a very, very loose grip on reality, so I appreciate your feedback. I do understand it's going to be very difficult to build the robot I'm talking about.

    I thought of using a motor instead of batteries, but I want to use the robot indoors as well. Wouldn't that be a problem?

    What kind of battery can supply power for such a robot for so many hours, and still be light enough for the robot to carry around? I'm thinking of using Li-Pos, because they're light. But it would take too many of them to keep the robot running for 4+ hours - I'll come back to the 4+ hours later. For starters, I think I will have to accept a much shorter uptime.

    Design-wise, I'm thinking of making a quadruped with small wheels, like skateboard wheels, instead of paws. That way, the robot will be able to drive around on flat surfaces, and still be able to tackle a few steps. Or at least the curve of the pavement. (Baby steps...)

    I believe that it is possible to build the kind of robot I'm talking about. Design, kinematics, electronics, frame, construction: it's not going to be easy, but I believe these can be done. I expect its "mind", the software that will make it do what I want it do, is going to be very difficult. But first, I'll have to build the thing.

    It's very, very hard for me to say :-) but I don't expect to succeed where entire University teams have failed. Not on my own, at least. At some point, I do intend to approach a university or two for help. And I do hope the community here won't mind a question from me every now and then :-)

    @Adrenalynn:
    Yes, it's going to be expensive. But I won't buy everything all at once. This is going to be a long project. I'll start small and work towards the robot I'm talking about.

  7. #7

    Re: I, Gömböc

    Quote Originally Posted by Gorbag View Post
    fact, building a series of robots each of which emphasizes a few of your desires will keep you occupied for many years, and you'd be learning valuable lessons in engineering which can carry forward to more ambitious projects in the future.
    That's definitely my game plan.

    Quote Originally Posted by Gorbag View Post
    Consider a radioactive tag if short range, (lead suit costs extra)
    I would never have thought of that one

    I like the tether / leash idea. I'm not so sure I'd like carrying 100lbs of batteries on my back though

    Everybody, I get it. I really do. I just wanted to get the requirements out there. As I said, I like to know what it is I'm working towards.

  8. #8

    Re: I, Gömböc

    Quote Originally Posted by Adrenalynn View Post
    I just noticed another design criteria. You want a tortoise that can climb stairs. I've had various turtles in my life and the only way they navigated stairs was in the jaws of my Jack Russel TurtleTerminator... I'm wondering about the purpose of that requirement?
    The Gömböc is a mono-monostatic object. It's self righting. I thought that the shape of a Gömböc would be good for a robot that has to move around. That way, if it falls on its back, it'll be easy for it to right itself.

    Quote Originally Posted by Adrenalynn View Post
    I The only way you're carrying that kind of weight around, especially on stairs, is with WAY bigger servos.

    I wish I were a supermodel. Alas, there comes a point when we realize that we have to accept basic universal truths. One relates to microservos carrying multiple pounds of payload, and the other relates to me being a supermodel. And neither are pretty - literally and figuratively.
    I get it. I'll be happy if the first version can carry my keys around for about a quarter of an hour. Could you suggest a WAY bigger servo?
    Last edited by Michalis; 04-26-2009 at 04:13 AM. Reason: To add a question.

  9. #9
    Join Date
    May 2008
    Posts
    2,228
    Images
    155
    Rep Power
    125

    Re: I, Gömböc

    Lipo will have the highest power to weight ratio, I don't think any other chemistry is gonna get you as close to what you want.

    As for servos, something like the Dynamixel RX-64 or EX-106 would be more in line with what you are looking for, or possibly the larger Vantec ones. Note, each of these servos will set you back $200-500.. ouch

    For inspiration on a tortoise with wheels see:
    [ame="http://www.youtube.com/watch?v=MFgePgQ73Ws&feature=related"]YouTube - Quadruped Robot[/ame]

    -Fergs

  10. #10

    Re: I, Gömböc

    lnxfergy, thanks for the tip - I didn't know the Vantec servos. Also, that's a beautiful quadruped! It has many of the features I want for my robot.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •