[clmazzac]

Problem

The IMU reported base_link acceleration wherever it was mounted. On a rigid body, a point at offset r from base_link also has two extra terms:

• tangential: α × r, from changing yaw rate
• centripetal: ω × (ω × r), from turning

Measured on main in a steady 0.5 rad/s turn at 3 m/s, with the IMU 1 m ahead and 0.5 m left: the IMU read 0 m/s² forward and 1.5 m/s² left, the same as at base_link. Physics gives -0.25 and 1.375.

Fix

• captureVehicleSnapshot adds bodyAngularAcceleration, the yaw-rate change since the previous snapshot.
• New accelerationAtOffset(snapshot, offset) adds both terms with Three.js Vector3.crossVectors.
• buildImuMeasurement uses it with the IMU mount position (sensorConfig.pose.position) before gravity and the sensor rotation.

Checked with a local script, outside the repo, IMU at (1, 0.5, 0):

Case	IMU reading (forward, left)	Expected
Steady 0.5 rad/s turn	-0.250, 1.375	-0.250, 1.375
Yaw rate rising at 0.2 rad/s²	-0.110, 0.495	-0.110, 0.495

Effect on existing runs

The default IMU mount is at base_link, so default runs do not change. Only runs with an offset IMU mount see new values.

Tests

• npm test on a clean copy of main with this and the other open sensor PRs merged: 1968 pass, 6 skipped, 1 fail. The failure is Python decodes retained RGB, depth, and CameraInfo SFLog messages, which also fails on main.
• No unit test is added in this PR.

🤖 Generated with Claude Code