Interaction Fit: when the robot's behavioral model does not match the environment.

A robot designed for warehouse aisles behaves differently in a clinical hallway. Interaction Fit measures the gap between where a robot was designed to operate and where it actually does.

Interaction Fit is not about whether the robot can physically navigate a new environment — most can. It measures whether the robot's behavioral assumptions match the social norms, spatial rhythms, and human expectations of the deployment site. Stopping distance, approach angle, speed modulation, and proximity tolerance are all calibrated for a context. When that context changes, the calibration does not automatically follow.

A robot calibrated for 2.4-meter warehouse aisles, redeployed in a clinical corridor at the same approach speed, reads as aggressive to staff and alarming to patients. The robot is not broken. Its behavioral model was authored for a different environment. The mismatch is Interaction Fit failure — not a technical fault, and not something a firmware update resolves without deliberate re-calibration against the new context.

The distinction between low Interaction Fit and a broken robot matters for how deployers respond. A broken robot gets repaired or replaced. A low-Interaction-Fit robot gets tolerated — complained about, worked around, and eventually sidelined — because the people experiencing it cannot name what is wrong. "It makes people uncomfortable" is not a service ticket. It is an Interaction Fit audit finding.

Deployers moving a robot between environments — even similar ones — should treat the transition as a new deployment, not a relocation. That means re-running the RXD Interaction Fit assessment for the receiving context, re-calibrating behavioral parameters, and running a structured observation period before declaring operational readiness. The machine moved. Its behavioral model did not move with it automatically.