I’m clearly missing something because although I can see the experimental proof that two wheels is better (see: Stryke Force), I don’t understand why that is. Help?
Sent from my iPhone using Tapatalk
Execution. It’s not about the swerve, it’s about the performance of the team as a whole.
After all, statistically, WCD is far more successful (see: everybody). Yet swerve seems to perform better on paper than a WCD, and a properly done swerve should beat a WCD in pure performance.
It’s all about execution. And although it’s possible to do swerve and build a good robot, it’s more probable to not do swerve and build a good robot. In a mere 6 weeks+, no one has enough resources to build a perfect robot, and the resources put into making a swerve that’s may/may not be better than a WCD might be better used in improving another part of the robot.
Strkye Force has better execution that most other swerves, and have spent years of iteration lowering the resource cost during build season.
Anyhow, a sample size of 1 is not exactly significant.
In their paper, they say “Overall height and complexity were reduced relative to previous iterations, reliability was improved, and the maintenance load on the pit crew was dramatically reduced.”
The dual wheels allow the bevel to be on the same shaft as the wheel, reducing height of the module and the complexity of having multiple gear trains on the rotating part.
It also introduces problems, like volume used and causing the robot to tend to shift around while the modules are rotating.
Iteration and testing are the only real answers, and what’s right for 2767 may not be right for you. Two wheels aren’t better, they’re a tradeoff, like everything else.
As Chak said “execution”.
I would put 40% of their success on the driver. He is amazing. It takes a special human to drive swerve and take advantage it.
Thanks everyone! From watching FOC I can tell that their driver is gifted. Interesting how their controls are set up, with one joystick for rotation and one for field oriented direction. Maybe this gave him a leg up?
Sent from my iPhone using Tapatalk
1 Like
Their code probably has a large element to it as well. In 10 seasons of watching swerve drives with great drivers, I have never seen the roll-off maneuvers that 2767 pulled off this season.
I would be very hesitant to say that every decision made by a well-tuned swerve is necessarily optimal in all cases. Just because Stryke Force had an awesome swerve drive doesn’t mean every choice and tradeoff they made was the right one, or was the one that produced their competitive advantage.
It looks like in this case, the 2 wheel design was for packaging purposes (to make a shorter module by putting the bevel on the wheel shaft while still keeping power transmission coaxial). The tradeoff was increased turning scrub among other things. All design choices involve tradeoffs and cost / benefit analysis; this particular game’s incentive for large ball hoppers probably pushed them toward shorter modules.
The 2-wheel design with the bevel gear in the center is very cheap and very low-profile to make; it provides some of the benefits of bevel-beside-wheel swerves while being much easier to manufacture. These benefits come at the cost of adding way more scrub when turning individual modules and having a much wider module, at least where the wheels are. Mainly for space constraint reasons I’m not a big fan of the style, but there are situations that can make it worth it. As always, I recommend doing swerve in the offseason long before you want to do it in-season.
Does anyone have a link to some of these maneuvers?
I don’t know about any specific matches, but I was some interesting driving in their highlight video.
They even pick-and-roll the poofs at one point (ish). There should also be a few examples in their IRI elims matches.
In general, if you’re looking for something to match with your own swerve, a reasonable goal is the translate-and-rotate maneuvers. The smoothness of 2767’s translate-and-rotate is impressive. So is the fact that the robot continues moving in a straight line during a translate-and-rotate maneuver.
The trick is some voodoo which handles a changing center of mass that causes different tractive forces on each individual wheel as they move positions while the robot rotates.
Thanks!
Looking at this, the first half the the maneuver (the quarter turn from wide to long as it’s passing 1986) is pretty similar to something I’ve envisioned in my head for a while - but seeing just how smooth that is on an actual robot is something else. I wonder if it can be improved by moving the axis of rotation from the center of the robot to the corner that (from the driver’s perspective) moves from front left to rear left; essentially creating a sort of ‘swing’ motion before being combined with the diagonal translation. I imagine it would be beneficial, but I’ve never actually seen it done for confirmation.
The second half - the quarter turn from wide back to long - is something clever yet simple that I hadn’t thought of. It moves the rear end away from the defender while moving front end in front of it, essentially completing the pass in a way that’s very difficult to stop. Again, I wonder if a rotation that resembles a swinging motion would improve it a bit.
If I were the driver, though, I would definitely change the control setup for rotation, as I feel the conventional right stick x-axis for rotation is too imprecise for these maneuvers (major props to their driver, as I’m pretty sure that’s what they’re doing). I’d prefer something like what’s described in Ether’s Halo-AR setup (the gist is that the direction of the robot follows the direction the stick is pointed) - but splitting the translation and rotation between the left and right stick. This gives position-based control of the rotation, which I find is much easier to do high-precision maneuvers with. (While I never used such a scheme in a competition, I added it as an experiment to our 2015 bot after the season. After tuning it to work somewhat well, I already greatly preferred it to the conventional setup I used in competitions.)