Between perception, models and actions, what does it mean for a robot to understand?
A robot can detect an obstacle, choose a route and adjust its speed. But does it understand what it is doing?
In robotics, we design systems that perceive their environment through sensors, process data and act according to rules or learned models. These capabilities can be impressive.
Yet completing a task does not automatically answer the question of understanding. Depending on what we mean by the word, we might refer to a representation of the world, reasoning, intention or subjective experience.
An open question
Avoiding an obstacle may result from a simple rule or a complex model.
A robot can adapt to new situations without that proving it is conscious.
The concepts of understanding and consciousness remain debated in philosophy and research.
What interests me is designing useful machines while being precise about what they can do — and what we cannot yet explain.
