Configure your drive motors in MecanumDrive as explained here. Make sure to properly reverse them using MecanumDirectionDebugger by following the official docs.
Tuning currently also requires a properly configured Control Hub or Expansion Hub IMU in MecanumDrive. This will be fixed in the future, but for now make sure your hub orientation is properly defined.
Also, make sure to configure the OTOS in your hardware config. By default, SparkFunOTOSDrive will look for a sensor named sensor_otos, but you can change this in SparkFunOTOSDrive line 70.
Note that, to mitigate an issue with the OTOS driver in SDK version 9.2 and to allow for driver modifications, you must currently configure the OTOS as "SparkFunOTOS Corrected" in your hardware config.
First, tune the Angular Scalar by running the OTOSAngularScalar OpMode and following the instructions. This will allow you to get the maximum accuracy from the OTOS IMU.
After you have tuned the angular scalar, the IMU will be accurate enough to tune the heading offset. Run OTOSHeadingOffsetTuner and follow the instructions.
Next, tune the Linear Scalar using the LocalizationTest OpMode. Again, ensure that you perform this tuning on field tiles so that the OTOS gives accurate data. SparkFun's official instructions to do are as follows:
To calibrate the linear scalar, move the robot a known distance and measure the error; do this multiple times at multiple speeds to get an average, then set the linear scalar to the inverse of the error. For example, if you move the robot 100 inches and the sensor reports 103 inches, set the linear scalar to 100/103 = 0.971
Set the Linear Scalar at line 56 of SparkFunOTOSDrive.
Next, tune the Position Offset using OTOSPositionOffsetTuner by following the instructions. This should ensure that the OTOS is properly aware of its location on the robot.
As a final step to ensure everything is working properly, run LocalizationTest again and try driving around, spinning in place, etc. to make sure everything is working properly.
Follow the official RoadRunner docs for the remaining tuning steps. However, make sure to start at ForwardRampLogger and follow the steps labeled "dead wheels."
Additionally, when you run AngularRampLogger, you can ignore the odometry position graphs and only use the trackWidthTicks one.
Once you have completed the official docs for tuning, you should be good to go to use Roadrunner as normal!