Blog Post with GIFs of it working - http://blog.spectrum3847.org/2018/01/day-16-virtual-4-bar-intake-clips.html
I have been looking forward to seeing this thing in action. It looks impressive thanks for sharing.
Love it, especially the use of a virtual 4-bar.
Do you have plans for automating the ‘grab’ action?
If we end up going with this design yes, automating the grab is in on the list either with a proximity sensor or 2, or an actual push plate switch at the back of the claw.
For the benefit of those of us rusty on their 4-bars*: can you explain what makes it a virtual 4-bar?
*This is one of those days where I totally hide behind my marketing degree.
There is technically 1 bar, but its able to keep the manipulator in the same orientation like a 4 bar. I am pretty sure Spectrum is doing this with belts. It’s illustrated really well in 95’s 2015 arm video.
There are two belts on each arm that hold the wheel plates perpendicular to the main plate with the motors attached. These belts aren’t driven by a motor or anything they just react to the motion of the swing arms. Giving them the motion of a parallel 4 bar that many people are used to seeing on arms. The idea for intake was based on 1538s actual 4 bar intake from 2015 with the goal of making it more compact to fit on an arm.
A 4-bar linkage has, well, 4 bars. A ‘vritual’ 4-bar uses two tensile members and one compression member to replace two opposed bars of the traditional 4-bar linkage. I.e. the two white or two black links in this picture:
Spectrum has used two pulleys of the same tooth count to maintain orientation of their collector arms. Paring pulleys of different tooth counts can achieve different types of motion.
Do you have any more pictures of how the cylinder is set up for actuating it? I am having a hard time seeing how you are getting that much rotation out of a cylinder.
There is a cylinder, its just not used for the rotation shown in the gif above.
What happens when a cube goes in such that a corner is close to the center of the mechanism? It seems like it would slip out.
The cylinder is driving the belt directly. Inspired by 971’s defense manipulators in 2016. The prototype is a 75mm stroke cylinder and 36t 5mm pulleys, so we get around 150 degrees of a rotation from it. The real version would likely be less stroke, the arms don’t need to fold back that far. Or we could stop having them tied together and have independent cylinders drive the arms directly.
It’s not a perfect intake, rarely any of them are but it’s doing better than expected at some of our tests.
I believe the key is in the slippery backstop. As the wheels provide roughly equal torque about their axles, if the corner hits the backstop a bit off center, the one farther away will provide greater torque than the other around the corner which hit the backstop due to the longer moment arm. This will cause the cube to rotate so that the intersection moves farther away. This greater torque cascades until the cube is lined up. If the strike were at the dead spot, vibration will likely cause it to cascade one way or the other; if not, rotating the robot a bit should flip it one way or the other.
I’ve really liked following your build blog, has been interesting seeing the design come along.
I’ll be curious to see if you find a good way to intake at an angle or if it will be simpler to just mount the intake so it is close to horizontal at ground level. With our prototypes we found the re-orienting ability of the horizontal rollers to drop quickly with increased angle so we had to have a cube lined up close to straight to grab it once the intake was rotated 45 degrees, and we just shaved down our compliant wheels if the cube came in corner first. Our test rig was much cruder though so you may have better results.