A point can be numerically valid and physically wrong when its coordinate frame is unclear.
Robotics systems commonly connect sensor frames, the robot base, odometry, and the global map. A detection from a camera cannot be used as a navigation goal until the transformation chain is known and time-aligned.
The frame chain
The map frame represents the global environment. Odometry provides locally continuous motion. The base frame represents the robot body. Sensors add their own offsets and orientations.
TF makes those relationships queryable, but only if each transform has one owner and timestamps remain compatible.
Debugging order
Before tuning a planner, inspect the frame tree, static transforms, timestamp freshness, and pose consistency. A planner cannot compensate for a goal expressed in the wrong frame.
This is why architecture diagrams for robotics should include coordinate frames and transform ownership, not only boxes labelled camera, model, and robot.