This contains a test procedure that you can use to characterize your drivetrain so that you can more accurately control and simulate it.
The test procedure involves running your robot in autonomous mode several times while the data logger program gathers data. It is important to stop autonomous mode when the robot is about to hit something, as the programmed autonomous mode will NOT stop your robot automatically.
Sample robot code is available in several different variations.
Python robot code also has RobotPy physics support, so you can run the tests in simulation if you want to see what the autonomous mode does! Unfortunately, it does not model battery voltage realistically, so you will only get a kv value from the simulation.
Required to use the data_logger.py/data_analyzer.py scripts.
- Install Python 3.6 on your data gathering computer that will be connected to the robot's network
- Once finished, install matplotlib, scipy, and statsmodels
On Windows the command to install matplotlib and statsmodels is as follows:
py -3 -m pip install matplotlib scipy statsmodels
Your robot must have encoders attached to the drivetrain to measure the robot's velocity.
If using a Python robot program, see the RobotPy installation documentation to install software needed to deploy robot code, and how to install RobotPy on your robot.
If using a Java robot program, see the WPILIb screensteps documentation for installing the necessary software on your computer.
Preparation: make sure the code won't make your robot go crazy
- Select one of the included robot programs, modify it to reflect your robot's drivetrain and encoder settings
- Deploy the code to your robot
- Ensure that pressing forward on a joystick makes your robot drive forward
- If not, modify the 'inverted' flags in the code, repeat steps 2/3/4 until the robot drives forward
- Open SmartDashboard/Shuffleboard/OutlineViewer and ensure that the l_encoder and r_encoder values are incrementing positively the correct distance in feet when you push the robot forward
Now you're ready to characterize your robot! On your data gathering computer, launch data_logger.py (you can double-click it on Windows). Enter in your team number or robot IP address when prompted.
Once the data logger has indicated that it has connected, do as prompted. Here's what it will prompt you to do:
- Slow driving forward: Enable robot in autonomous mode, disable before the robot hits something
- Slow driving backward: Enable robot in autonomous mode, disable before the robot hits something
- Fast driving forward: Enable robot in autonomous mode, disable before the robot hits something
- Fast driving backward: Enable robot in autonomous mode, disable before the robot hits something
Once you have ran the 4 autonomous modes, the data files will be written to this directory and a window should pop up that plots the data for you.
If using RobotPy and the pyfrc robot simulator, you can plug the generated kv/ka/vintercept parameters into the TankModel for your robot for a more realistic simulation experience.
Refer to the original drivetrain characterization paper for ideas on how to use these empirically measured constants to improve control of your drivetrain.
Finally, help the rest of the FRC community! Post your raw json data + drivetrain characteristics at https://www.chiefdelphi.com/forums/showthread.php?t=161539
This is intended to be a project that all members of the FIRST community can quickly and easily contribute to. If you find a bug, or have an idea that you think others can use:
- Fork this git repository to your github account
- Create your feature branch (
git checkout -b my-new-feature) - Commit your changes (
git commit -am 'Add some feature') - Push to the branch (
git push -u origin my-new-feature) - Create new Pull Request on github
All code in this repository is available under the Apache v2 license.
Dustin Spicuzza (dustin@virtualroadside.com)