Core structure

MARSLib is built using a “Hardware Abstraction Layer” (like a universal remote control) that lets you swap out motors without rewriting your code. We built the framework to reuse memory instead of constantly creating new objects (called “zero-allocation”), which stops the robot from lagging. Because of this, whether you are running a real robot or a simulated one, the math behaves the exact same way.

structure Diagram

    
%%{init: {'themeVariables': { 'edgeLabelBackground':'#111111', 'tertiaryTextColor':'#ffffff' } } }%%
flowchart TD
  Z["User Project<br/>(RobotContainer.java)"] -->|Depends On| A("MARSLib Framework<br/>(Vendor Dependency)")
  A --> C[Swerve Subsystem]
  A --> D[Mechanisms]
  A --> E[System Utilities]
  C <--> F{"IO Abstraction Layer"}
  D <--> F
  E <--> F
  F <-->|Hardware Mode| G["IOReal<br/>(TalonFX, Pigeon2)"]
  F <-->|Simulation Mode| H["IOSim<br/>(Dyn4j Physics)"]
  G --> I[("AdvantageKit Log")]
  H --> I[("AdvantageKit Log")]
  style Z fill:#0a0a0a,stroke:#2a2a2a,stroke-width:2px,color:#e8e8e8
  style A fill:#141414,stroke:#B32416,stroke-width:2px,color:#e8e8e8
  style C fill:#1a1a1a,stroke:#2a2a2a,stroke-width:1px,color:#e8e8e8
  style D fill:#1a1a1a,stroke:#2a2a2a,stroke-width:1px,color:#e8e8e8
  style E fill:#1a1a1a,stroke:#2a2a2a,stroke-width:1px,color:#e8e8e8
  style F fill:#B32416,stroke:#d42e1e,stroke-width:2px,color:#fff
  style G fill:#2a2a2a,stroke:#1a1a1a,stroke-width:1px,color:#29b6f6
  style H fill:#2a2a2a,stroke:#1a1a1a,stroke-width:1px,color:#9c7bcc
  style I fill:#0a0a0a,stroke:#6ba3d6,stroke-width:2px,color:#6ba3d6

Package Reference

Core framework packages and their responsibilities.

Package	Description	Key Classes
`com.marslib.swerve`	250Hz odometry, swerve kinematics, traction control	`SwerveDrive`, `SwerveModule`, `PhoenixOdometryThread`
`com.marslib.vision`	AprilTag localization with MegaTag 2.0 fusion	`MARSVision`, `VisionIO`
`com.marslib.mechanisms`	Generic IO abstractions	`RotaryMechanismIO`, `LinearMechanismIO`
`com.marslib.faults`	Thread-safe fault reporting	`MARSFaultManager`, `Alert`
`com.marslib.simulation`	Dyn4j physics world, game piece spawning	`MARSPhysicsWorld`, `GamePieceBody`
`com.marslib.power`	Voltage load-shedding and stator monitoring	`MARSPowerManager`
`com.marslib.util`	State machines, tunable numbers, shot math	`MARSStateMachine`, `EliteShooterMath`
`com.marslib.auto`	PathPlanner integration, alignment commands	`MARSAutoBuilder`
`frc.robot`	Competition logic: RobotContainer, commands	`RobotContainer`, `MARSSuperstructure`

📖 Further Reading & External Resources

Limelight MegaTag 2.0 Breakdown - Understanding how FRC 1690 style IMU yaw-seeding rejects rapid AprilTag noise variations.
PhotonVision Hardware Guide - Best practices for illuminating targets globally.