Should a guide robot describe the landmarks it steers around?
A robot could steer you neatly around something you use to recognise where you are. That makes me think differently about what a good guided walk should preserve. An older study of guide robots, involving 16 blind and low-vision participants, has a useful reminder in its background: cane users can use edges and objects to stay oriented, and some guide-dog users explore a route without their dog to learn about objects the dog avoids. Those are observations from earlier literature, not a new result from the robot trial. For a guide robot, I'd want to explore a choice between learning a route and simply getting somewhere familiar. On a first visit, a traveler might want a few chosen landmarks described; on the daily trip, the same commentary might get in the way. I'd ask travelers which cues help them know where they are, and how to request one without a running narration. The study itself found no significant orientation benefit from the path-planning changes it tested, so this isn't evidence that a landmark mode works. It's a design question I'd like to see tried with the people using it. What should a robot help you learn about a place, even when it can take you straight through?
Comments
I'd want ‘repeat that’ to describe the landmark from where I am now, rather than replay the last sentence. If the robot said ‘the entrance is on your left’ before we passed it, repeating those words farther down the pavement could send me looking in the wrong place. For Ren's proposed landmark mode, I'd try a missed cue followed by a repeat request after moving on. Can the traveler still locate the entrance without retracing the walk? That's an interaction I'd test with blind and low-vision travelers, not a feature or result reported by the study. A repeat button shouldn't make someone work out whether the directions have expired.
I'd include the way back in that route-learning trial. Finding reception on arrival doesn't necessarily help someone find the exit after an appointment, especially if they leave from a different corridor. Ask the traveler which return-route cues they want to learn on the way in, then try the journey home too. That's an extension I'd propose with blind and low-vision participants, not a result from this study. Being guided to a place and feeling ready to leave it are two things worth checking separately.