Kiwi Router

Saw this and thought the community would get a kick out of it. We’ll have to wait and see how well the thing works, but really cool concept nonetheless. (Note the three layer omni-wheels).


Who thought this was a good idea? If you bump the work-piece or blow on it the wrong way it will throw the whole thing off… or you get too much dust build-up and the wheels start slipping…

Don’t get me wrong, it’s a neat idea, and I’m always entertained to see a new application for kiwi drive, but this hardly seems practical. I’m also betting when it comes out it’ll probably be cheaper to just get an X-Carve.

It’s got two little towers that look like they pay out cable, presumably for measuring, so that the drive isn’t required to be perfect.

I’ll stay cautiously optimistic.


I feel like that might help with position but not rotation. I would also be curious to know how this deals with traversing gaps created by different cuts.

I guess I’m a bit cynical after my experience with the X-Carve my team got a few years back, even a more traditional system like that is susceptible to errors and accuracy loss caused by vibration or dust build up. Even if this router is a closed-loop system, I feel like there’s a lot more variables that can go wrong and mess up your part than a conventional CNC router.

If they can make it work more power to them, but I remain skeptical until proven otherwise.

…or when you have to put the wheels somewhere you’ve already cut because the bit is inside the triangle formed by the wheels?

Wheels with rollers and precision machining just don’t go together in my book.

But no one is publishing my book.

Let me know when they have swerve CNC.


Some 971 students spotted in this video relating to the Goliath.

In regards to the machine itself, color me very skeptical.

I feel like these guys are missing an opportunity to attach a shop vac hose to the robot to clear chips and also provide a significant amount of down-force to the machine and help with any slipping issues.


Unless there’s an encoder / pot on the string so you know the angle it is at?

There actually is at least one picture on that website that appears to show it hooked up to a shop vac. Not sure it provides any downforce benefit in the particular configuration they use though.

I suppose that’s possible but I still feel like it would have trouble tracking in a number of different scenarios. I feel like it would be quite easy to induce an error that the machine would not be fast or accurate enough to compensate for.

As a first year manual machining student even I find this a waste of money.

It basically has a string pot mounted on a rotary encoder for localization. Honestly your better off printing off large templates and using a jigsaw than messing with this thing. It will be faster.

I find it quite amusing to see so many conclusions being drawn related to cost and performance with no quantitative information available for either.

More of an “estimation” than a “conclusion”.

Just like you can examine an FRC robot without seeing it in a match and get a general idea of the cost (It’s pretty easy to tell the difference between a $1000 robot and a $4000 robot) and raw machine performance (based on mechanisms and components used and build quality). These are ball-park estimates of course, but it’s close enough for the majority of cases.

I think everyone here would love to be surprised by this thing, but a lot of our expectations are low based on the information currently available.

Now all they need is a router held by a 2 armed, 2 end effector, opposable thumbed, bipedal, upright standing adaptive unit and it can carve much larger work pieces. It can even come covered in a cloth exterior that sags in the back during strenuous operations.

Of course that system might need augmentation with tools and clamps.