Robots in public space must follow human-like etiquette

What it is

When a robot moves through a corridor, an elevator, a queue, or a sidewalk, it is expected to follow the same etiquette conventions people use in those spaces: not cutting into the middle of a line, yielding in a narrow passage, keeping clearance when passing behind someone. This is not a cosmetic layer of politeness bolted onto navigation — it decides whether bystanders treat the robot as a normal participant in the local traffic order or as an obstacle that needs constant watching. The practical test is concrete: do local pedestrians extend to the robot the same short-horizon prediction ("it will do what a person would do here") they extend to each other.

Why it happens

The convention holds for robots because bystanders' interpretation of movement in public space is already scripted before they know what is approaching. Seeing any body near a queue or a narrow passage triggers the default script for "how anything moving through this scene should behave," and that script fires regardless of whether the mover turns out to be a person or a machine. A robot entering such a space is therefore already inside the scope of that script whether or not its designers intended it to be. When the robot's actual path breaks the script — cutting straight through a waiting line, taking a passage a person would have yielded — bystanders do not suspend judgment because "it's just a machine that doesn't know better." They react with the same irritation and wariness they would show a rule-breaking person. What this shows is that the trigger for the negative reaction is the violation of the local traffic script itself; the robot's non-human identity only affects whether the violation later gets attributed to malfunction rather than rudeness, not how sharp the in-the-moment reaction is.

Studying it

Etiquette violations are poorly captured by surveys or video ratings alone, because the core phenomenon is an immediate behavioral reaction under real crowding and time pressure — remove that pressure and what gets measured is attitude, not behavior. The common design is to deploy a robot performing queuing, corridor-crossing, or narrow-passage tasks in a real or semi-real public setting, manipulate whether it follows human conventions (queue order, yield timing, how early it slows down), and measure near-miss counts, detour behavior, bystanders' negative-affect ratings, and whether they later comply with a request from the robot. Direct interactants and passive bystanders need to be logged separately, since their tolerance thresholds typically differ, and field studies need to record crowd density and composition, because expected yield timing itself shifts with how crowded the space is.

Where it stops holding

Etiquette expectations belong to the bystander, not to the robot as a physical requirement, so purely functional spaces without bystanders — a warehouse floor, an unstaffed corridor, a kitchen pass-through — do not need this constraint. "Human-like" also does not mean copying every human motion detail: wheelchair users, someone carrying a child, or emergency staff already hold different priority in human traffic norms, and a robot that applies one uniform human convention can miss exactly these distinctions; a robot can also signal intent through lights, projection, or on-screen text in ways a body cannot, and this channel is not bound by the "act human" constraint at all. Robots whose form or motion differs sharply from a person's — a low cleaning robot, a delivery cart that reads as pure equipment — also draw a lower yielding expectation from bystanders than a human-scale or humanoid robot would, so the same strictness of convention cannot be applied uniformly across robot forms.

Applying it

Path planning and avoidance strategy should carry an explicit soft constraint for local traffic convention on top of efficiency and safety terms: staying at the back of a queue rather than merging in from the side, slowing early in a narrow passage to leave room to yield, widening lateral clearance when passing behind a stationary group. These parameters should come from observing the target deployment site rather than a single global default, since expected yield timing already differs across venues — a hospital corridor, a mall, a transit platform. To verify, run field observation comparing a convention-following path strategy against one that ignores it, and count negative-reaction events (frowns, detours, verbal complaints) under each; that frequency, not the path planner's own efficiency metric, is the relevant outcome.

Related

• Same group: X6.06.2 Interrupting or cutting in line provokes disapproval much like it would from a person · X6.06.3 Etiquette expectations vary by culture and setting, with no universal script · X6.06.4 Violating etiquette expectations erodes bystander trust in a robot's later behavior
• Nearby: X5.02 Navigating shared space · X2.01 Signaling intent before acting
• Search terms: robot etiquette · social navigation · proxemics · human-robot interaction