Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

The maximum speed with this gearing configuration should be about 14 fps right? Is that Nighthawks's speed?

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

Your team has been kind enough to share the design of your swerve and describe how it works when I have visited in your pits. When thinking about it, I have always been unsure about a manufacturing aspects of your design.

How do you get alignment of the "tower" to the wheel support structure? How do you assure center to center alignment? and then how do you assure perpendicularity of the tower to its mating piece at the bottom. Is it just a matter of being very careful with conventional tooling, do you somehow machine it as one assembly or is there some other hidden trick you can reveal?

I would think this would be particularly important, doubly so because of your integral, custom bearing assembly.

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

The only thrust bearing system is machined in the plate that houses the integral part of the swerve module, so that is what I will assume you are asking for. The thrust bearing is machined into two opposing plates and smashed together by the screws that also happen to hold it together. Here is a picture to show the inside of what it looks like:



(I circled the two channels for the ball bearings. Also, that sprocket is NOT floating, it is being held up by a piece of delrin that cannot be seen in the part section view.)
The two black plates sandwich the BBs between them and act as the thrust bearing.

I felt the need to add pictures of the plates themselves. The smaller round one is hidden just above the blue 3D printed part that is an unmentioned experiment for this year.





The maximum calculated speed is close to 14 fps, but the real speed due to outside forces such as friction is closer to 12-13 fps.

The method of attaching the "tower" as named above to the plate itself is actually much simpler than you might think. The thrust bearing actually is what holds the module and sandwiches the mounting plate between it and a couple of machined pieces of delrin on top. The piece is just a small aluminum block machined into an angled bracket that is also what compresses the thrust bearing and holds the entire assembly together. All of these parts are machined with CNC mills (some manual mill work is added but nothing major) to ensure all the parts will fit together very tightly and accurately, providing the perfect 90 degree angle between the module and the plate. The image below has the small plates (only one is visible) highlighted in red and the 3" tube highlighted in green for ease of visibility. More images of the overall assembly can be shown if needed to get a full perspective but until then I don't feel the need to clog up others' bandwidth too much with all these pictures!



If you have any other questions just let me know!

-John Taylor Novak

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

Thank you for such a complete response regarding information on your swerve module design and manufacture.

We tried swerve a few years ago and while we eventually got it to mostly work, it co-opted some of the time that perhaps we should of spent on other subsystems. We haven't completely given up on it, but it certainly isn't a favored approach right now.

As implied in the thread above, the software and control part of this takes work as well.

There is a kind of elegance to pieces of your design that I admire.

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

I'm curious about the slip ring. Can you share the manufacturer and part number?

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

I can send (and have sent) a private message to anyone wondering what the slip ring model is.

-John Taylor Novak

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

Hello JTN. i am Tristan Shepard (aka Skippy), I am a student and am part of our pit crew with (kyEOT) or (Engineers of Tomorrow team 2783). we had talked to your pit crew and some of your mentors ate the Queen City Regional, we talked to you all about how you made your swerve. They said if we wanted the CAD we just had to contacted you all. we used the wild swerve from andymark but saw a lot that could be improved on it. I was wondering if our team could get the CAD for your swerve module that your team used. Thanks. Skippy

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

Why did you use 6 inch wheels over 4 inch ones? As far as I can tell, I think that 4 inch wheels would make the entire assembly smaller and lighter while needing less gear reduction. Any reason why I'm wrong?

How are the grooves for the thrust bearing cut?

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

I was wondering about that same thing. I've always been a fan of small wheels, especially when designing a swerve drive.
I have couple more questions too:
1. What's that pneumatic cylinder doing in your CAD render!? It really caught me off guard.
2. What is the experimental 3D printed part below your thrust bearing?

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

As the wheels do not have encoders, this could have some issues with going straight for moderately long distances. I think that the pneumatic cylinder and the notches in the blue piece might be an experimental solution. I'm sure someone on the team can answer this more accurately.

After playing around with some math and CAD, I realized that smaller wheels mean smaller reductions, which either means that the intermediary shaft has to get closer to the CIM, or the sprocket on the wheel has to get bigger and closer to the ground. I think that a 24/60 gear reduction with a 16/40 chain run will fit, but just barely. (EDIT: 24/64 and 16/38 fits better). The smaller wheel also reduces the space for the thrust bearing, but I think I can get it, or a thinner solution, to fit. These issues might be fixable by a first stage belt like 192 has recently used, but I have no experience with these.

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

Could you clarify where this cylinder is? I can't seem to find it in any of the CAD screenshots that have been posted.

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

I had the chance this past weekend to see their swerve pods in-person, including the experimental one with the 3D printed blue plate. I'd like to thank the members of Team 16 for answering all my questions with patience.

In the picture posted, the piston is located under the black mounting plate, and adjacent to the blue one (its the grey cylindrical object in the cad). From what they told me, it was designed to test out and eliminate the twitching that the modules may have when trying to drive in a straight line. Because the control system tries to compensate the drift, and then over-corrects the action, the piston would lock in to the appropriate groove when needed and make sure the robot drives in a straight line.

They did mention that they would have to do some more testing with the current design since the motor was able to overcome the piston lock fairly easily.

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

In some ways that sound like a cool solution, and thinking outside the box. But it also sounds like they're fixing a software problem with hardware.
Regardless of my mixed feelings about this. I'd be very interested to hear if it helped or not.

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

We have found 6" wheels to be more suitable because they have more tread touching the carpet (realistically) and therefore more traction in most scenarios. The grooves were cut using a simple ball end mill... pretty easy to figure out if you ask me.

The pneumatic cylinder was used (as described above) to lock the wheels in place only during wheel calibration autonomous to avoid the need to use any sort of software to keep the wheels straight and move in a straight line. It is much easier to just not supply power to the rotating modules and let them remain physically straight than run a control loop to correct it, slowing us down and removing accuracy. We didn't implement this or further develop it because of the lack of any need in this game. The 3D printed part contains the grooves that the cylinder shaft deploys into.

This is entirely wrong. We didn't even USE it this year and there was no problem with control. I'm not sure why you would think a team that has been using swerve for a decade wouldn't know how to go in a straight line, all that is necessary to achieve that is to push forward on the throttle.. As stated before, the main purpose was to ensure consistent calibration (which we only do about once per competition) and to make going in straight lines in autonomous easier due to the lack of reliance on a gyro that it would produce.

-John Taylor Novak

P.S. As of now the CAD models have not yet been released to the public but I can supply images of the model instead.

Re: Baxter Bomb Squad 2014 ~ Nighthawk
 

The Inherent problem with the gearboxes we can afford (weight as well as price) is backlash. The resolution and control should be sufficient to hold a precise position but mopping up deadband is very difficult. Some attempts include raising the gain to the point of oscillation to average out the slop. This small amount of error hasn't been noticeable for driving in teleop but can cause significant drift when driving across the field autonomously.

The cylinder was intended to improve two issues.
1. We currently calibrate the wheel positions by rotating them all straight and running a task to save the voltage from the encoders as the reference values. These are saved to a file and used as the offset from then on. The cylinders would allow more precise mechanical positioning for that.
2. We would use them to ensure home position at the start of the match. notches around the perimeter of the bearing plate could allow accurate re-positioning at 45 deg or 90 deg or other intervals.

The cylinders are not intended to force the module to a position. when extended, the module software would oscillate at low energy around the setpoint until the cylinder is seated in the detent. This would be detected by a cylinder mounted limit switch on each wheel.

Obviously this was not required for autonomous success this year. We are always trying to mop up limitations of the hardware and software for the next time precision is needed.