Gyros and IMUs: Pigeon 2.0, NavX2, and ADIS

A gyroscope measures angular rate (degrees per second). Integrate that rate over time and you get an angle. An IMU (Inertial Measurement Unit) combines gyros with accelerometers (and sometimes a magnetometer) to give a fuller picture of orientation: yaw, pitch, and roll.

Yaw, pitch, roll

Using aviation conventions: yaw is rotation about the vertical axis (which way the robot faces — the value you care about most), pitch is tilt forward/back, and roll is tilt side to side. For most FRC driving you use yaw; pitch/roll matter for balancing or detecting tipping.

Drift — the core challenge

Because angle comes from integrating rate, tiny rate errors accumulate into drift: the reported heading slowly wanders even when the robot is still. Common sources of drift include sensor bias and temperature change as the device warms up. You manage drift by:

• Choosing a modern, low-drift IMU.
• Following the manufacturer's recommended setup (some IMUs sample a zero-rate bias at boot and need stillness; others, like the Pigeon 2.0, do not).
• Correcting heading periodically with absolute references (like AprilTags).

The popular FRC IMUs

• CTRE Pigeon 2.0 (CTR Electronics): a 9-degrees-of-freedom IMU on the CAN bus. Unlike the original Pigeon, it requires no on-boot calibration and no temperature calibration, and it does not need to be held still at boot — you get useful heading as soon as it powers on, with dramatically reduced drift. The only optional step is a one-time mount calibration in Phoenix Tuner X once placement is final. Read heading in Phoenix 6 with getYaw() or getRotation2d(). Its yaw is continuous (passes 360 to 361, not back to 0) and increases counter-clockwise viewed from the top (NWU convention for Rotation2d).
• Kauai Labs NavX2 (navX2-MXP / navX2-Micro): a popular IMU that mounts to the roboRIO MXP port (SPI) or connects via USB/I2C, providing a fused yaw via its onboard AHRS.
• Analog Devices ADIS16470 / ADIS16448: three-axis IMUs that connect over SPI (typically via the MXP); WPILib ships first-party classes for them.

Reading a gyro in WPILib

All of these expose a heading you can wrap in a Rotation2d. Convention check: WPILib expects counter-clockwise-positive yaw, so make sure your gyro's sign matches (invert if needed). A continuous, CCW-positive heading is what odometry and pose estimation expect.