Currently my team uses solidworks for all our cad. This year we tried to run simulations using gazibo and the plugin for solidworks but we couldn’t get it to work. We are going to be trying out synthesis in the off season now though. Is there any way to get SW files into inventor and synthesis without losing constraints data? We don’t want to switch to inventor but we will if needed. What does your team do regarding simulation? Is it absolutely necessary?
What problem are you trying to solve with simulation? Are you trying to determine if your mechanical/motor choices will work? Are you trying to test your software?
Simulation is not necessary to be successful. If you are familiar with SW then switching to inventor just to get simulation working probably will hurt you more than it will help you.
In my experience, simulating by exporting CAD is more effort than it’s worth. We do simulation in code, by creating mathematical models of how we would expect the system to respond (based on the numbers for mass and moment of inertia that we get from the CAD). This lets us run code for specific mechanisms, and shows us, given a certain input (voltage, usually) what the output is (angle, distance, etc). We’ve found this more useful than simulation by exporting CAD would be.
Our goal is to give drivers some practice while the robot is being built and to allow the programmers to test some of their code.
Makes sense.
For driver practice, I wouldn’t think that simulation would be the best way to practice (although I’ve never actually done a full robot simulation before). Driving a past year’s robot, even if it’s speed/etc is slightly different would probably give someone more useful practice than driving a simulated robot.
For programming, I’ve found that there are very effective ways to test code without relying on CAD (except for measurements like mass and MOI). A lot of the logic for state machines/etc can be tested without any simulation at all. For our actual controllers, we tend to use state space controllers, which make running simulation easy (since finding a model for the system is a prerequisite for writing the actual controller). Austin from 971 has given some good talks about testing code which you can find at Fall Workshops - Citrus Circuits (look for “Test Driven Development”. “System Modeling” might also be of interest). We’ll also be releasing our code for this year soon, which has tests written for all of our mechanisms.
We’ve found that we can do drive practice and code testing very effectively without using simulations like Gazebo. If you have any questions about how we do simulation feel free to ask. Or if you do want to use Gazebo/etc, that’s fine as well, although I won’t be able to help with it.