Mr. Ehnert, Lathrop High School's engineering department Instructor and LHS FIRST Robotics club Coach, has 15 FIRST Tech Challenge (FTC) teams that he takes to the Alaska State tournament every year. (That's a lot of nuts and bolts).
I decided to make a project that I could leave with Lathrop as a sort of "going away" present for Mr. Ehnert. That's when the idea of a "Magic Sorter" was born. The purpose of this Tetrix Nut/Bolt Sorter was to decrease the amount of time spent on sorting robot parts. By doing so, it would also increase the time for productive robot building. In the class, I made 3D printed and laser cut models to solve the sorting problem and complete it's purpose. However, the "Magic Sorter" didn't work as well as I envisioned. So, I worked on improving it over the summer.
I started playing around with mechanisms in my Lego box. And, well I decided to build the whole thing out of Legos. I didn't have a "blog" back then, and I didn't take any pictures. The pictures below are from when I came back during Winter Break after my first semester at MIT.
|Hopper and Laser Sorting Section|
Here's the Tetrix Sorter in action! I took a bunch of short clips when I came home for Winter Break. I finally got around to compiling them into a "pretty" video. Enjoy!
That last clip is at a weird angle. I wish I could take a better video to get the whole thing functioning, but I'm currently about 5238 kilometers away from it (thanks to Wolfram for that statistic).
I like to make the NXT securely attached to my robots. Though, this sometimes has consequences. Such as accessing the USB port and the battery pack. I would have to use a flat head screw diver to pry the battery pack off.
The motor above the NXT controls the hopper conveyor belt at the top. There are four ninety degree turns, through three bevel gear pairs and one worm gear (to make it very slow), along its path to the conveyor belt axis. There is actually a hidden axle going up through the left and right main support posts.
The light sensor at the top of the hopper is actually not used for sensing light. It is used as a lamp to indicate when the hopper is running. Thus when the hopper was jammed, I could prevent the motor from being unhappy. (Sometimes, when building up the hopper, screws would get caught in places screws shouldn't go).
My biggest challenge during the PLTW EDD project was the hopper. I tried a bunch of different methods to isolate the screws and dispense them one by one. (I tried a bin with a grooved wheel at the bottom which would spin and dispense the screws out a small hole. This ended up jamming all the time. Another method I tried was a vibrating table. This worked; however, it was unpleasantly loud). Over the summer, I came up with the idea of a slanted conveyor belt. The slant is the key part, because it keeps a bunch of screws from dropping off all at once. Also I added a flap at the top to help knock the piles of screws down to just a few.
The hands in the picture above belong to Aven Bross. He was the programmer on my robotics team (3705) for the 2011 and 2012 seasons. He helped a ton by programming the Tetrix Sorter right before I left for MIT in the fall. He also helped fine tune the Tetrix Sorter while I was away. Thanks Aven!
Finally, a picture of the whole Tetrix Sorter.
I designed the sorter to separate the two types of screws, button head and socket cap screws. The little blue Lego piece, that the servo pushes the screw/nut on, has a gap with a critical distance. It is big enough for socket cap screws to fall through, and lets the button head screws slide on top. Then the bin can be separated into two compartments with button and socket cap screws sorted (in theory).
Here is a close up picture of the servos. I used a four bar linkage on the servos to push the nuts and bolts off the conveyor belt.
|Sorted... For the most part|
The sorting containers click in with one pin in order to be easily removable.
|Sorted bins :)|
For the longest time, the light sensor wasn't as accurate as I would've liked it to have been. I added another light sensor to increase the chance it read accurately. By doing this, I learned that the problem with the accuracy of the light sensor was the positioning.
It turned out that the laser was "leaking" around the screws. I adjusted the height of the light sensor and put black paper to enclose the sensor from ambient light. This significantly improved the light sensor readings to within 1 raw value each time the same part passed.
Since I buried the NXT under the hopper, I needed a bunch of long cables to run from the servos and NXT conveyor belt motor back to the NXT. There weren't any NXT wires long enough for my application so I used NXT to RCX adapter cables with a super long RCX cable in between. However, because I went through the adapter, I lost all the sensor readings from the NXT motor (because the rotation sensor goes through the NXT wires unlike the separated rotation sensor with the beloved yellow RCX system).
I began searching for long NXT wires on the internet and learned that there was a good reason for not making NXT wires longer than 2 or 3 feet. When they are that long, the signals induce signals into the other wires and mess up the sensor readings from the motor.
So I made my own wire with the individual wires separated (by soldering the separated wires to two NXT ends).
I sandwiched the wires between two pieces of duct tape.
I pinned through the duct tape wire with some long Lego pins to secure it against the servo posts.
My inspiration for the Lego version of the Tetrix Sorter:
If you figure out how they oriented the blocks before scanning them for color, please let me know.