Also consider that Air Traffic Control and marine VHF radio[1] (and anything else based on old school radio) are a "standalone sound interface". It requires a formalized setting and training, sure. Rattling off grid coordinates isn't exactly relaxing, but it's a thing.

For usability, consider something more along the lines of "Hey roomba, go clean up the spill in the kitchen. [...] You missed a spot near the door."

[1]: As in maritime, not military.