Who needs mecanum wheels?

Skills are all that’s needed to solve a problem. Take this four-wheeled robot as an example. [Michal Zalewski] wanted it to be omnidirectional but wasn’t very satisfied with the concept of mecanum wheels and the like. So he designed a chassis with wheels at each corner that can pivot as one to change orientation. The image may look like a rendering at first glance, but this is actually the physical prototype. See what we mean about skills?

Okay, so the robot design is pretty cool. But we’re more excited about the build process. We’ve looked at [Michal's] work before. He wrote a thorough guide about CNC mold making. These parts are all cast from epoxy. This starts with a rough milled mold, which is given a second pass for the fine details before being painted with a release agent and used to make a silicone mold. From this the parts are produced. Check out the Flickr set showing the casting process for the planetary gear box on each motor. If only these results were as easy to achieve as he makes it look.

[via Reddit]


  1. Vince says:

    Just wow!

  2. jonzilla9000 says:

    I just killed an hour on that guide to machining and mold-making. Absolutely fantastic.

  3. fightcube says:

    Holy crap Batman! This is some fine engineering and craftsmanship here!!!

    BTW: This steering mechanism may already be patented, but I know for sure the wheel concept is patented by iRobot for their PackBots.


  4. SpydaMonky says:


  5. HAD says:

    Would have been significantly faster and cheaper to just 3d printed it using SLA or polyjet.

    • g19fanatic says:

      maybe… but no where as cool nor precise (to within 3 mils)…

      • Hackerspacer says:

        16 microns is not accurate enough?

        16 microns = 0.0006299 in

      • Cyril says:

        @Hackerspacer LOL. Go on then…

      • g19fanatic says:

        @Hackerspacer, I’d like to see a 3D printer get 16 microns AND still be faster or cheaper than this method… UV Epoxies are not exactly cheap… nor that fast when you have to mold the dozens of parts it takes for just ONE of these gear sets…

      • Hackerspacer says:

        You wouldn’t be making MOLDS – you would print the parts directly. No milling, degassing, silicone molding, degassing, casting, curing, removing flash, accounting for shrinkage on the epoxy and silicones – you get the picture. Print it in a UV epoxy, heck even a filled UV cured epoxy. Done.

        Method 1)
        Come up with STL files -> accounting for shrinkage on the epoxy and silicones -> getting tooling -> milling -> degassing -> silicone molding -> degassing -> casting -> curing -> removing flash -> measuring parts -> done.

        Method 2)
        Come up with STL files -> Print. Done.

        Don’t get me wrong. There is serious artistry here. But you can use another technology and get much quicker and more accurate results – possibly even cheaper as well.

      • g19fanatic says:

        you still haven’t come up with a current example of something that works at better than 3 mils, is cheaper than molding (which is very cheap), is as fast (you make the molds ONCE and with care you can reuse to your hearts content… need to make 4 planetary gear sets? each with a dozen or so pieces? no problem… make one molding and you can then make 4 sets without issue…)

        There is a reason that 3D printing still isn’t the main method of plastic forming… it isn’t as cheap nor as fast. I’d dare say even for one offs, not just pieces profiting from the economy of scale.

        oh and don’t tell me that your Method 2 is that simple… It takes TONS of tinkering to get 3D prints the way you want them before you can pump them out… Just the same as when you’re molding… it all becomes part of the process.

      • Cyril says:

        Still haven’t explained where you’re getting an honest 16 micron resolution from…. or anything like it either ;)

      • Michal Zalewski says:

        Hackerspacer: I see the sentiment for 3D printing, but the question of accuracy aside, that comparison of workflows is pretty bogus.

        First of all, you’re comparing apples to oranges: you can directly machine parts on CNC mills (more so than print them, given the greater choice of engineering-grade materials), but there are clear benefits to moldmaking, and a reason why people buy 3D printers just to make molds. Heck, there are even wax printers that aren’t good for anything else, and they sell for big bucks.

        The details of the workflow are also largely bogus. I don’t understand why “measuring parts”, degassing a master mold, or “getting tooling” are on your list. There is usually no practical shrinkage you need to account for in room-temperature polyurethane resins, too: it’s not injection molding.

        Consequently, for 3D printing, you usually have multiple additional steps involved, depending on the technology: removing support structures or melting out filler wax, sandblasting the surface, post-curing the resin in SLA systems, etc.

        And really – there are probably no sub-$50k 3D printers that can get the detail and accuracy of a $2k mill. Even if they claim a high resolution in X-Y, there’s usually some fine print about the Z axis, or the minimum size of a surface feature they can print. In a couple of years, this will probably change; wax-printing systems are pretty promising, in particular.

    • asdf says:

      Faster than using his CNC to just cut the parts straight out of an easily machinable plastic?

      Correct me if I’m wrong, but he chose this method only because he could easily/quickly mold additional units.

  6. RandyKC says:

    I want to be like you when I grow up!! You can tell when someone takes pride and adds a little artistry in their work.
    Thank you also for your guides that you have generously taken a lot of time to write. I was just getting ready to bite the bullet and start forking out some cash. Your guides have probably saved me a lot of money and misery.
    Please keep us posted!

  7. Rudager says:

    It will be interesting to see this move with fluid movements rather than: Stop… straighten wheels… move forward… stop… rotate wheels 90 degrees… rotate machine. If it’s even possible.

    Very nice.

  8. Gear Pr0n says:

    This is pretty cool, but i found his cycloid drive even more interesting (warning: gear porn ahead!) http://lcamtuf.coredump.cx/cycloid/

  9. Robot says:

    To my eye, this level of work blurs the line between engineering and artistry in a way that is rare and pleasing to encounter. This hacker’s respect for process is admirable. I hope to see some complex segmented joints or something like that in the future.

    – Robot

  10. Jeff says:

    amazing build. kudos to michael and thanks for sharing.

  11. AB says:

    Very cool to see how to go from concept to making it with that precision and the documentation of the process. That´s a direction i want to go as well… one day.

  12. Jerry Ficke says:

    The mars rovers use a similar navigating scheme except this one takes it a step further and can just crab over sideways. Very cool.

  13. Keith says:

    Impressive work! I thought it was funny how he shows the miniscule costs of the materials to make the parts after making molds with a $21K milling machine!

  14. mark says:

    Really nice. I would go ahead and add the other two drive motors though if it were me.

  15. spiralbrain says:

    Very impressive!
    I wonder if the middle motor can be replaced with a servo perhaps. Looking for a video of this in action but didn’t find any. The build is very well documented and the work has so much precision.

  16. steve says:

    Unbelievable. Really great stuff and much better than omnidirectional wheels.

  17. asdf says:

    His reasons for not doing an additive process:


    Look under “2. Milling? That’s so old school!”

  18. Alex says:

    Definitely one of the best projects I’ve seen on HaD in a long time. Clever idea, and very well executed.

  19. Jerry says:

    Just absolutely wow! Precise engineering and art in one design. I am a techie, but have worked with artists, and the combination produces some amazing results. This guy is both an engineer AND an artist at the same time, a very powerful combination. I wish you long life and prosperity kind sir, and thank you for sharing with the public!

  20. Jack says:

    While it is beautifully crafted, I suspect that there is no video of it in motion because that would expose a very significant design flaw.

    When the wheels are turned to go forward, the two drive wheels drive in the same direction. When the wheels are turned 90 degrees to go left, the drive wheels again drive in the same direction to go left. If the wheels are put in the 45 degree position in between, turning the drive wheels in opposite directions will let it rotate…. but that’s it. There are no other drive configurations that allow it to not be dragging the idle wheels sideways.

    If the idle wheels had turned the opposite way, so all 4 wheels were always pointing in the same direction, then it would lose its symmetry, but at least be able to drive in all directions and turn while in motion. I think that would be a crab drive.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


Get every new post delivered to your Inbox.

Join 97,511 other followers