ludov

23. Box Factory Completely useless, but highly fascinating. The "factory" consists of a 4-step sequencer that actuates: The box being pushed into the handler The raising of the handler The rotation of the handler The box being pushed out into the handler Gravity then helps the box back to the first step of the sequence. Here's a movie to show that it actually works, and some detail photos (high-resolution is available on my Bricksafe folder).

Damnit. I spent some time tidying up the hoses, reinforcing the structure, etc., and guess what: nothing works anymore. Not even the sequencer, which I didn't touch. Part of the problem seems to be the compressor, which is too weak, despite using three pumps. I guess that they're either too old and/or they have too much air to pump around, but I'm puzzled why the compressor worked fine yesterday and won't move anything today. With two hand-pumps it sometimes works, so maybe that can be my backup plan. But with just one day left to get everything else to work again, I doubt I can manage.

Holy smokes it actually works: Obviously, I have some hoses to clean up and general improvement of some parts of the structure, but heck, it actually works. As you can see, I changed a lot with respect to the initial "dinosaur" prototype: it turned out that it was way too complicated to move 2 cylinders in one step, in time. So I came up with this "trebuchet"-like contraption, which works quite reliable. Like any other good contraption, the boxes occasionally get stuck, but it's quite reliable in general. One cylinder of the sequencer is really slow (I guess it's one of the oldest in my collection). I might change it, but since it is the one after the raising of the arm, it is a bit convenient as it give the arm time to come up to the top. My little old compressor pumps struggle to make it all go, so hence the delay in pushing the box out. What do you guys think?

I didn't think of the raising of the undercarriage. It would indeed be inconvenient to push on what you're trying to raise. Don't get me wrong though. also now it looks great! You got my vote at least

I had to write it down with some truth tables, but it makes perfect sense now. I got stuck after adding dedicated switches and hadn't thought of using an additional valve to control the supply air to those switches. Genius! I might borrow this for my box factory, if you don't mind (although, maybe I should focus on finishing it first, rather than thinking of enhancements ). Thanks for the explanation; you have my vote!

It looks very nice! I'm curious to know how you implemented the sequence "up, push, retract, down", i.e. "extendA, extendB, retractB,retractA". The normal sequence (as in 8868 B-model) is "extendA, extendB, retractA, retractB". I found out that you can't just swap some hoses to reverse an order, since you end up having air pushing into both ports of a cylinder (if that description makes any sense).

Awesome! These machines are great and yours is a great rendition. Very well done! The only thing I can comment is maybe the pump: it's a bit awkward in its current location, in my opinion; perhaps you could camouflage it as an exhaust or something?

Very nice, I like this a lot! Aside from the bucket maybe, I think it is a bit out of proportion, but that's up to you. What's the little blue pump doing down there? What are you showing by disconnecting it? Nevermind, it seems I missed a post.

There's the expert! I'm actually using the old switches, because they are indeed easier to move. Also using the old cylinders. Maybe I should check all for leaks, as some of them are quite old (my first pneumatic was from http://brickset.com/sets/8042-1/Multi-Model-Pneumatic-Set). What's your opinion on mixing "logic air" and "work air"? By the way, I was first considering to include one of your AND or OR gates, but figured that would be a bit too much for me at this moment...

I saw it in the background of one of your photo's. It is pretty to look at; if only I had a shelf to put it on, I'd rebuild it as well...

Actually, they are difficult to get in, also in my opinion.They seem more stubborn than regular pins; not sure why. I've been following this topic silently, but I sure do enjoy it. Thanks for sharing your efforts Grum, especially the 8880, that was a lot of memories coming back.

We had this discussion before, and maybe before that...

Wasn't this part introduced together with 8880? I recall it was used together with this part to make the steering rack: There the collar was needed, obviously.

This makes perfect sense! Learned something today

I use the 3/4 pin when the half-liftarm needs to rotate. But often I also find myself just use the 1/2 pin for the same reasons. However, it seems that if you first put the short end in the half liftarm then it goes in too far, and the other long end won't snap anymore (does anyone understand this? ). I'm not sure which one TLG prefers to use, but for this latter reason I would assume they'd use the 3/4 pin.

Ok, this turns out to be tricker than I anticipated. I built a compressor with three of these 5.5L pumps. When I use two of them to run the machine and one to run the sequencer, the machine moves pretty quick, but the sequencer won't move unless I help it. The other way around though, using two pumps to run the sequencer and one to move the machine turns the table around: the machine hardly moves, and the sequencer goes through its stages really quickly. I have two of these 6L pumps, so I could build another compressor, but in any case my problem remains: the sequencer goes to its next step way too fast. Any ideas how to approach this? Am I aiming too high? (I'm trying to figure out how 8868 B-model solved this; it only had one compressor so obviously was mixing work and logic air. It also had only two steps, but using three cylinders. I need to study this more...)

Yes, I intend to have two completely separate circuits. Especially moving the arm up takes more pressure than the sequencer needs, so I'm anticipating that if I don't separate the two circuits the sequencer will toggle before the arm moves.

When TC10 was announced, I just had to come up with something. I didn't collect all those pneumatics for nothing...! However, due to work, I had very little time until now, but that gave me plenty of time to think about what to build when I would have time. I already saw a lot of great entries, and it was hard to come up with something original. It should have a sequencer, because I never built one before, except for the B-model of 8868 Air Tech Claw Rig. After much contemplation, I came up with the following (apologies for the crappy photo: it's late and the light is far from optimal): This is not a dinosaur. It is my prototype box factory. It will move boxes as shown in the front-right (2x2x2 studs) in a GBC-like perpetual motion. Or, that is the idea at least. In the back is a 4-step sequencer. Each cylinder toggles two switches, so that I can separate "logic air" from "work air". The contraption on the right is my "robot". The arm moves up and down, the whole thing can slide forward and backwards on the stands, and the "head", which contains the box, can rotate 90 degrees. The long 1x11 cylinder hanging from the back of the sequencer is going to push a box in the "head", and the small one is going to push it out again. So far, everything sort of works. The big challenge is to make sure that the robot completes his motions within one step of the sequencer. I plan to do this by putting a multi-pump compressor with one or two air tanks on the "work air" channel and a "weak", single pump compressor on the "logic air" channel. I hope that the beefy compressor will make the motions of the robot quick enough. The prototyping with two hand pumps seems to indicate it's feasible. So, I need to build two compressors, a guiding channel for the boxes to bring them from the dumping point back to the beginning, and prettify the whole thing. Four weeks to go, right? What do you think? Can it work? Any advice?

This are for me also the reasons why I don't like video reviews. The only parts where video reviews add value (for me) is to show how things move. In e.g. Sariel's reviews I skip everything until the "Functions" part. I think a voice over would be useful there, because then you don't need to read while looking at the mechanism at the same time (I often find myself watching sections twice: once to read the explanation and once to see the mechanism move). I like how Jim does his reviews: mostly text and pictures and a few videos to show how things move where it adds value. No voice over needed, as those video clips rarely exceed 20 seconds and they are presented in context. In this way I can consume my reviews in silence

Great, that was very informative! You're approach is simple yet effective. Plus I learned a new algorithm, so that's good

I'm particularly interested in the two nonlinear optimisations that you're doing (for the camera calibration and the trajectory estimation). I've always been interested in nonlinear problems, but never had much chance to do anything cool with nonlinear optimisation. I can make an educated guess on your approach based on what you've told so far, but I'd love to see the math and theory behind it. I'm out of academia for a while now and in either case I wasn't to much into vision problems, but the two approaches appear very innovative to me. Are they original ideas from you or adaptations of existing theory?

This is really awesome. As a robotics engineer, I can really appreciate the effort and the result. Thanks for sharing, I wouldn't mind a more in-depth explanation of the algorithms if you're willing to write it up!

Very nice indeed! The 379 is a favorite. Any pics of the drivetrain? Looking forward to the video!

Definitely manual. It would fit better with the rest of the model that is already manual. Also, I wonder if two function (forward/backward and recline) for two seats, so 4 power functions, can be elegantly integrated. Ok, not meant to doubt your skill but you already indicate it's kinda busy in the chassis. If you want to use the battery box for a motorised function, how about adding an idler motor somehow? Using a differential, I would imagine the V12 to idle while the model is stationary and you'd "add" the forward speed to it (through the gearbox?) when you move the car.

That video is awesome! And the model too I have no words