This years Robot code features many improvements over the previously design action system, IO classes, and provides a much nicer developer experience.
Actions are no longer an interface, and instead are now an abstract class. This allows for a robust system that eliminates the annoying need for "firstRun" variables and complicated task creation. This system has been majorly improved upon with a new event based system with full Action life cycle events. These include beforeFirstRun(), afterFirstRun(), afterFinished(), and of course the main event onRun(). For more detailed information on the new event system, see Creating Actions.
With the new actions there is of course a revamped way to preprogram autonomous strategies. After analyzing the code from last year we realized that it was just clunky and not easyily maintainable. We attribute this to the large use of chained if and else if statements, and dozens of methods to assemble autnomous programs.
The new system merges everything into one convenient location for programming autonomous strategies. in the frc.robot.autonomous package there is a new enum called Auto. Here you can easily assemble programs by passing in a display name, optionally a starting position on the field, and a argument vector of Actions. Using the position selector on SmartDashboard, we can automatically filter through the programmed autonomous strategies and only show those that can run in the selected Position. This allows us to remove all the nasty if statements.
To assist new members with learning our codebase, I will be creating some basic guides/tutorials about how to use parts of the code so that everyone understands how the new systems work.
robotInit, teleopPeriodic, etc
During the robot's autonomous control phase (the first 15 seconds of the game), the drive team is not able to control the robot. Instead, we have to create a preprogrammed route (or usually a lot of preprogrammed routes) in order to instruct the robot to move without our input. These instructions can be simple like "wait 5 seconds" and "drive straight 1 meter", or complicated like "aim towards the goal and fire a projectile". Reguardless, we need a way to give the robot a list of reusable instructions to tell it what to do during autonomous. This is were Actions come in.
An Action is just a task the robot is programmed to do, that can be measured as being complete or incomplete. For example, a task can be "wait 5 seconds", and we can measure its completeness by seeing how much time has passed since the task was launched.
to be continued...
- Create custom scripting language for actions programming
- Add pneumatics controls
- Add camera input
- Add PWM motor controlers
- Add joysticks back