1 Autonomous Systems
-
When systems must work in dangerous environments where humans cannot be nearby, and so humans cannot assess the possibilities easily and quickly;
-
Similarly, systems that must work in remote environments where direct human control is infeasible;
-
Situations where systems need to react much more quickly than humans can possibly achieve;
-
Scenarios where, while human control may be possible, there are just too many autonomous entities active for any one human to keep track of; or (increasingly);
-
Where it is cheaper to use autonomous systems rather than involving a human pilot/driver/controller!
2 Assessing Autonomy
2.1 Hybrid Agent Architectures
-
Assess, and possibly revise, the information held,
-
Generate new motivations or revise current ones, and
-
Decide what to do, i.e. deliberate over its motivations and information.
2.2 Hybrid Agent Architectures
3 Towards Genericity
can a common “autonomy architecture” be provided and be configured for different autonomous systems?
-
Easier to develop (through re-use of code/architecture),
-
More reliable (since the particular core would be deeply analyzed and refined over several configurations), and
-
Easier to deploy on new platforms.