|
|
|
Organizations / United States / SMART / David Schilling / CrateContraption / 2006 / Crossroads | |
| The Crate Contraption - Crossroads Shopping Center 2006 Now that we, after several years, have all the kinks worked out of our Crate Contraption (as if!) we decided that we could complicate things by adding a Great Ball Contraption to the display we put on once or twice a year. This allowed a lot more people to participate. While we’ve gotten much faster at it each year, an issue we’ve had With the Crate Contraptions is the amount of time it has taken to integrate all the components, and get them working together. One of the nice things about the GBC is that people can just bring whatever they’ve put together, and with minimal effort (that is, no more than an hour or so) there’s a hugely complex ‘device’ that people can enjoy. Right off the top, I want to thank everyone who participated! We had more modules than space! At the end of the day the GBC had processed exactly 4321 balls! (One of Mark Kenworthy’s modules counted each ball as it went past.) Pictures of the contraption can be found here. Videos are DIVX encoded. They can be found here:
The Layout The primary problem we had to solve with a combination of the Crate Contraption and a Great Ball Contraption was the inherent difference in the rate of ball movement. For the GBC to seem ‘alive’, it’s necessary to have approximately one ball per second make its way through each module. But the Crate Contraption throughput is closer to perhaps fifteen balls per minute, give or take. We could have had both just occupying the same space, and then it wouldn’t have been an issue, but that was a cop-out. Somehow balls had to leave the GBC and enter the Crate Contraption, and later had to be returned to the GBC. With the almost certainty that there would be a very long line of GBC modules, this meant at least one robot would have to make a long trip, and would mean an even slower rate of ball throughput. The solution was that there had to be multiple paths for the balls to return to the beginning of the GBC. At the end of the row of GBC modules we’d have to have one that split the balls so that most would just go back to the beginning, while a small amount would be siphoned off to go through the Crate Contraption. There would also have to be a module at the other end that would combine two streams of balls. That one was much easier to design. We decided to put the Crate Contraption at one end of the line of GBC modules. This was done because then only one robot would have to travel a long distance to deliver balls back into the GBC. Since we didn’t know how many modules there would be ahead of time, and since the number might theoretically have changed over the course of the day, this also made it easier, as only one robot’s track would need to be modified. We decided that the GBC would use only soccer balls, while the Crate Contraption would have a mixture of balls. Our final layout had a dump truck taking a load of soccer balls from the last module of the GBC, and dump them into a ball combiner that we used in one of our first Crate Contraptions. In the combiner were already a large number of basketballs. An NXT-powered scissor lift took the balls from the combiner, and dumped them into the sorter. The sorter filled a crate with soccer balls on one side. Gus’ Dumping Forklift took these and brought them back to the GBC. On the basketball side of the sorter was a contraption (originally to be a GBC module) that had a bucket lift, a ‘water wheel’ and a spinning sign that was powered by the water wheel. The wheel was actually powered by the basketballs. It dumped them into a trough, which was emptied by a front-end loader, and brought back to the combiner. One of the harder things to do was fine-tune the number of balls in this combined contraption. The Dumping Forklift used a pair of crates - one that it carried, and one that was waiting under the sorter to be filled. There couldn’t be too many soccer balls in the Crate Contraption part because then the waiting crate would get too full. So that meant the dump truck could only increase the soccer balls by an equal amount to that which were leaving the Crate Contraption part of the system. This was done by clever design of the last GBC module. It had a counter that would measure the number of balls going to the dump truck hopper. After the hopper was considered ‘full’, it would divert all balls to the train. So as to not starve the train though, even when the hopper for the dump truck was being filled, only one in four balls went that route. Even with this precaution, though, a lot more soccer balls ended up in the system than we liked, and over the course of the day we ended up removing a few handfuls every now and then, and returning them to the GBC line. The other aspect of the tuning was the number of basketballs - the water wheel was powered by them, so we needed an almost continuous stream of balls coming out of the sorter. This worked okay for a while, but then the motor and complex gear train on the bucket lift started wearing down, which slowed down the number of balls it could handle. So throughout the course of the day, we were constantly trying to adjust the flow of balls. The Robots Much of our effort went into the GBC end of the layout. So most of the robots were ones we’ve built and used in the past. But one was a special treat for everyone. Gus Jansson, one of the NXT beta testers was given permission by the LEGO Company to show off one of the NXT robots that he built ahead of the upcoming release date. It showed off some of the new features of the system. For instance, it didn’t just follow a line, like the other mobile robots did. There were two line segments that it used to line itself up nicely with its end points - the hopper that filled a crate, and the ball sorter that it dumped them into. But the rest of the journey made use of the built-in rotation sensors in the motors to drive in straight lines, and turn through a known number of degrees. The dumping mechanism was a clever use of mechanical parts put together into a scissor lift. Most clever was how it was designed to have a high amount of force available at the point where the scissor part is all folded up. This is usually the part that needs the greatest amount of force, but with a good lever arm, and the stronger NXT motors, this wasn’t a problem. Again, the only sensor used to control how high or low the lift went was the rotation sensor built into the motor! We can’t wait to try out some other NXT robots in our NXT Crate Contraption. Train GBC Modules There are a large number of GBC modules in the layout that I won’t describe. I encourage people who built them to post a write-up to our group. Here I’ll focus only on the two ends of the GBC line that we built to integrate the GBC with the Crate Contraption. As indicated before, we used a train to shuttle balls quickly back through the GBC line. Both modules used the same train-stopping mechanism. A wheel on a lever arm would constantly press a touch sensor down. When the train came along, it would press the wheel, and hence release the touch sensor. This would start the braking of the train, perhaps the toughest thing to do right. The stations were designed so that the train would just barely past the wheel, and the touch sensor then would be pressed in again, at which point the train needed to be stopped. If the train wasn’t stopped, the balls wouldn’t end up in the right place. If the train stopped too early, it wouldn’t get filled, or dumped. In that case we would pulse power to the train, slowly increasing the power if needed, to get it moving again. The train filling station had all the balls of the GBC come into it. At first 1/4 of them would be sent to a hopper that filled the dump truck of the Crate Contraption, while 3/4 of them would go to the train. When the dump truck’s hopper was ‘full’ (the balls were counted as they went there) then the rest of the balls would go to the train. The train itself was simply an 8-wide box car design with a side door that could be folded down. Using both rubber bands, and a magnet, the door would be held shut until it was pulled open. Each station was responsible for powering half the track. There was no communication between the two halves other than the train itself shuttling back and forth like a single ‘bit’ of information. We isolated the two halves of the track with tape at first, but after a few hours the train wheels cut through the tape, and suddenly both RCXs were powering both halves! The increased speed of the train caused a moment of astonishment, but the reason was quickly discovered. A better solution was suggested by someone to simply not have the two halves of the track touching at all. Put them a stud apart, and use plates to hold them in place. The momentum of the train was sufficient to carry it past this point, and the train still ran so smoothly, you couldn’t even tell that there was a gap in the track. So finally, the dumping module: it stopped the train, and a lever came down to open the door. The balls flowed onto a wide chain lift, and back into the GBC along a ramp. The ramp was sufficiently long to also allow the dumping forklift to dump its balls onto it. Thanks to Gus and David who built the robots in the crate contraption, and thanks to the following people who provided one or more GBC modules: Gus Jansson Mark Kenworthy Shane Brinkman-Davis Craig Zupke Wayne Hussey David Schilling Doug Bell Andrew S. Sean K. Patrick W. Alex K. Jennifer G. Lennart J. Tess R. David I. Rahul B. Gabriel P. Charles P. |
|
Primary content in this document is © David Schilling. All other text, images, or trademarks in this document are the intellectual property of their respective owners. |
| |
©2005 LUGNET. All rights reserved. - hosted by steinbruch.info GbR |
