Mobility results need localization context

Evidence and interpretation

REP 105 distinguishes a continuous but drifting odom reference from a map reference that may change discontinuously. This matters when comparing trajectories: a jump in reported coordinates can be a localization correction rather than physical motion. Maha’s proposed evidence record retains transforms and reference changes alongside the task outcome. Do not smooth away a discontinuity simply to make the trajectory visually convincing. The question is what moved, relative to which frame, and how the record lets another person determine that.

Proposed evidence workflow

1. Declare the reference frame for goals, observations and outcomes.
2. Retain localization resets and the transform history used in analysis.
3. Separate path completion from localization accuracy and human-interaction suitability.

Worked illustration

A target at x=4 in our grid is a cell in a fixed abstract frame. There is no drift and no localization algorithm. Comparing that result with a physical robot’s map coordinates without an additional measurement protocol would change the meaning of the result.

Limits

Neither a continuous trace nor a replayed map establishes safe locomotion. This page does not prescribe walking control, collision avoidance or human-proximity limits.

Sources and review

Continue reading

• Robot evidence needs explicit frames and units
• Replaying robot logs is not reproducing an experiment