Does anyone know of a physics simulator that could be used for testing drivetrain configurations? I’d like to be able to vary things like wheelbase, number of wheels, friction properties, center of gravity, etc, and see how the virtual robot turns. I realize nothing beats real world testing, but I feel like a lot could be learned with a decent simulator.
We had a fair bit of trouble with our drivetrain this year, and I’d like to try and learn more about why. It is a tall robot with a pretty standard 6WD configuration. We originally installed 6 inch pneumatic wheels, and the turning was awful, really bouncy and slow. Before our first event, we switched to colson wheels, which turned a lot better, but driving over the steel barriers was rough. We ended up breaking a bearing, which caused us to want to go back to the pneumatic wheels. I suspect our center of gravity is probably higher than it should be, although Solidworks claims it is roughly 12" from the floor which doesn’t seem too bad. Anyway, there are a lot of variables that go into a good drivetrain design, and if someone knows of a program I could use to test some of these virtually I’d love to hear about it.
You likely had too little center drop or the middle tire wasn’t inflated enough. That’s almost most likely what caused the issues with turning and the chassis jumping around
I feel quite the opposite. There are far too many variables to appropriately simulate friction interaction of a drivetrain in an efficient manner, some napkin math will tell you just about everything you can know without testing. As teams have discovered, just about everything influences how a robot handles; even our competition and practice robots had a tendency to feel different from day one.
You’re probably right. It’s been several years since we used pneumatic wheels and we likely didn’t have enough rocker. And the amount of rocker is another variable that directly influences driving performance that is very difficult to change once the drivetrain is built.
I agree that a simulation would probably not be very accurate, but having the ability to change parameters quickly and see how it influences the system would be really nice. I did come across Gazebo, which seems tailor made for simulating mobile robots, but it runs on Ubuntu which is a little beyond me.
Would this napkin math you refer to help in deciding the number of wheels, amount of rocker, etc?
A taller robot also has a larger moment of inertia in the pitch and roll axes. Even if the COG is low, this bigger moment of inertia can cause interesting feedback when you give certain drive commands.
I recall this being a huge problem for us in 2018 with the tall elevator robots and all traction (colson) wheels. If you rotate in place, given a certain nudge, the robot will oscillate uncontrollably, when the only input you give it is a constant turning command. This was fixed for us by using omnis in the corners to eliminate any scrub forces, and for some reason that worked. Not very useful for this year’s game with the metal barriers though.
I would be very impressed if somebody had a simulator that could capture these oddities, especially concerning pneumatic wheels. It’s one thing to make a simulator that spits out data, but modeling a complex system like this and spitting out data that actually matches reality is a whole different level.
It’s actually quite easy with pneumatic wheels, just inflate the middle wheels more and the corner wheels less
We used omni wheels last year for a similar reason. Our driver actually didn’t like how easily the robot turned! So omni wheels are a good option to salvage an otherwise poorly turning robot, but they break quite easily.
If you have a tall robot with traction wheels, I wonder if limiting speed during turns would help smooth things out.
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.