Autonomous
Build and test routines that run without driver input
Overview
An runs without driver input for the 30-second AUTO period of an match. It can read , move , and make decisions based on its program. The exact game tasks and scoring change each ; see the current season guide and current Competition Manual for those details.
Build an autonomous routine that repeats reliably, stops safely, and handles errors in localization and game-piece interaction. Adding steps can reduce reliability if each step is not tested.
Core concepts
- Motion planning describes where and how the should move.
- Localization estimates the robot's pose using , an , , or vision.
- Control uses sensor feedback to reduce the difference between the desired and measured behavior.
- Decision logic handles alternate outcomes, such as a missed game piece or a tag that was not detected.
These components affect each other. A path follower cannot compensate for a badly calibrated localizer, and an AprilTag pose is an estimate whose quality depends on visibility and camera calibration.
Learning order
- Test individual hardware safely.
- Make one short movement and verify direction, encoder direction, and stopping behavior.
- Measure localization error before adding long paths.
- Add one mechanism action and confirm its timing.
- Test repeated runs from the real starting position, with realistic battery and field conditions.
- Add fallback behavior and an explicit safe stop.
The following pages cover movement, odometry, path planning, and sensor fusion. They explain general methods; check each library's official documentation before using its API.