|
|
| |
OK, I had to post this in rtl because I confess the main inspiration for this
was watching Ian's amazing LEGO industrial P&P automations over the years. This
isn't nearly as fast as his, but it's the first time I've captured a motion I
really like, and I'm happy with the extensibility of it:
http://www.youtube.com/watch?v=-ZolqjEnhB4
OK, it could use a better name, but still. The other inspiration for this was a
contest I was in a while ago, to replicate a cube of 2x4 bricks. A lot of the
folks involved seemed to think that the lack of easy linear mechanism in the NXT
set was a significant problem. I didn't see it quite that way, at least not when
I had all these motor encoders "free" with the NXT, and decided to base mine
entirely on non-linear motion. I didn't quite finish it (gearing to push the
2x4's firmly together was a stumbling point), but really liked the manipulative
mechanism, so I improved it and started testing it. Since one of these "arms"
only needs two NXT motors, two NXT sets could drive three of these setups,
leading to a fun looking assembly line.
I'll toss some pics up on Brickshelf if anyone is interested, but really, this
is awfully simple - most folks here could replicate it from the video alone.
It's just fun to watch it work with that precision for a long period of time.
--
Brian Davis
|
|
| |
Hi Dino,
"samurai74" <baldi_nospam_@venus.it> writes:
> In lugnet.robotics, Dino Baldi wrote:
> >
> > I need your help concerning the vintage 1093 Interface A.
> > I found the 9750 version, with no interface cable and no software :-(
> >
> > Is possible to build-up a parallel cable for PC ?
probably yes, the interface is simply TTL I/O according to this site:
http://www.isodomos.com/Lego-Sets/1093.html
Note that I have no personal experience with this, use at your own risk.
Jürgen
--
Jürgen Stuber <juergen@jstuber.net>
http://www.jstuber.net/
|
|
| |
Nice work again Brian,
It is a nice action, is it accelerating? or fixed speed?, reminds me of a bird..
perhaps Peck'n'Place... :)
I noticed you have a calibration with a limit switch under the 'place' rack, are
your Pick racks also calibrated or fixed relative positions? if not maybe there
is room for another sensor on the base that rotates with the arm?
Also how are you detecting the last brick? it seems in the vid that it is using
an assumed number of bricks rather than detection, I noticed this when it
dropped the blue brick it did not move on but returned to the same rack.
Is this NXT-G still? (I know its your preference)
Its a very nice project I would be keen to see some closer pics/instructions
Regards,
Mike
|
|
| |
In lugnet.robotics, Mike Hatton wrote:
> Also how are you detecting the last brick? it seems in the vid that it is using
> an assumed number of bricks rather than detection, I noticed this when it
> dropped the blue brick it did not move on but returned to the same rack.
>
> Mike
Doh being blind it did move on...
|
|
| |
In lugnet.robotics, Mike Hatton wrote:
> It is a nice action, is it accelerating?
> or fixed speed?, reminds me of a bird...
Yes, the motion reminds me a a bird as well sometimes. The speed of the arm
motions is under program control - it will slow down smoothly as it approaches
the target position, for instance, and some movements are set to accelerate &
decelerate. Actually, it will move much faster, but at the cost of being
jerkier. I don't have it set up right now to accelerate smoothly up to the
midpoint, and then decelerate smoothly afterwords (no reason I couldn't I
guess), so instead each motion has a user-defined "maximum speed" that it will
not exceed.
> I noticed you have a calibration with a
> limit switch under the 'place' rack, are
> your Pick racks also calibrated or fixed
> relative positions?
In this example, fixed relative position. In fact, it works best if you do not
attach the "racks" or "pallets" to the main structure, but position each
independently. Tied together, it's very difficult to get the exact spherical
symmetry required by the arm motions (although possible - adjustable connection
lengths with axles are great for this). As to sensors, the ideal thing would be
a sensor on the "finger" itself, but I didn't worry about autodetecting each
station - as in practice, each station might be doing something completely
different, and therefore need special code. It's a good idea, though.
> Also how are you detecting the last brick?
Right now PnP will not sort black - as that's the color of the of the "finger"
when it's empty. So if the arm waves in front of the sensor and the only
detected color is black, it assumes the EOAT is empty, and advances the internal
pallet counter. When it advances the pallet counter too far, it moves back to
the first, and repeats.
> Is this NXT-G still? (I know its your preference)
Yes - because it's bloody easy for a lot of things, and I actually think it's
kind of fun*. Also, there were several minor irritations that spawned this in
it's current form - one of which was the large number of people that I've heard
complain about how "poor" the NXT or NXT-G is at reliably positioning motors. So
I wanted to limit myself to a minimal set of "tools", and show that's it's
possible... not even that hard, actually.
> Its a very nice project I would be keen to
> see some closer pics/instructions
I'll toss some images onto BS when I get the chance. I should probably LDD the
PnP arm itself, as it has so few parts it would be pretty quick to do... but I'm
not sure all the parts I need are in LDD (the quarter ovals, for example).
--
Brian Davis
*Yes, I know many will think that's a strange, twisted, or even perverse
statement... but considering that we spent our time engineering using only small
plastic toys as a constraint, "because it's fun" seems a valid motivation to me
:)
|
|
|
