Light, sound, and projection can supplement intent
Aliases: multimodal intent cue · robot external communication · explicit intent signal
What it is
Multimodal robot intent signalling uses light, sound, text, or floor projection when motion alone does not reveal goal, path, or state. Channels have different spatial reach: projection marks occupied ground, a directional light identifies a target, and sound captures attention outside the visual field.
Why it happens
Motion is constrained by viewpoint, occlusion, and path ambiguity. Explicit cues externalise an internal plan. Redundant modalities preserve meaning when one channel fails; complementary modalities require integration and can add load. All cues must remain synchronised with planning state or they display stale intent.
Studying it
Single and combined modalities can be compared for identical actions using intent recognition, detection distance, response time, false avoidance, workload, and accessibility. Redundancy benefit and semantic conflict need separate tests under measured illumination, noise, viewing angle, and floor material. Instruction-free trials avoid inflating public comprehension.
Where it stops holding
More channels do not guarantee clarity. Simultaneous light, speech, and projection can compete; sunlight washes out projection, noise masks sound, and colour-only codes exclude some observers. Signalling cannot compensate for unsafe motion, and unstable plans should not promise a precise route.
Applying it
- Assign one core meaning to start, path, yield, stop, and failure before choosing the modality suited to range and direction.
- Express critical safety meaning through at least two channels that do not fail under the same sensory condition; resolve conflict conservatively.
- Test first-use interpretation and erroneous action in target light, noise, occlusion, and traffic, synchronising cue logs with the active plan.