Question

Why do robots need Hidden Markov Models?

Answer 1

because they cannot observe state variables directly.

Answer 2

Hmm right I guess they need to account for their bad sensors :(

Answer 3

among other things.

Answer 4

namely, even if they sensors always work perfectly, they will nearly always still not tell you what the values are of the state variables you're interested in.

Answer 5

For example, the museum guide robot. It doesn't have a sensor that tells it where it is in the museum; it's (x,y) coordinates. It has sensors that tell it the geometry of objects around it.