Sliding Autonomy for a Human Centered System in a Space Exploration Domain

This thesis focuses on the topics of sliding autonomy and situation awareness.
The first addresses the capability of an agent to act at different levels of autonomy depending on internal and/or external contingencies. It is also possible to change the autonomy level during the execution as a result of previous accomplished tasks or directed by the human supervisor.
The main aim is to maximize the agent efficiency in facing unexpected situations and adapt its behavior according to the situation.
The second topic regards the amount and type of information the agent must provide to the human supervisor in order to let him/her be aware about what is going on. This helps the human supervisor to response to the agent's direct requests for help and able him/her to direct intervene anytime exploiting the human skills to help the agent to act at the best performances. This topic includes also the way in which the information are shown to the user, that means there is a preliminary study on possible graphical/augmented user interfaces.
The two topics are related each other in the sense that the more autonomy means (usually) the less information to the human supervisor and vice-versa, hence the autonomy level implies the quality and quantity of information to be sent.
During this thesis work I developed a prototype using a tool provided by NASA called PLEXIL, accompained with a task interpreter called Universal Executive for executing the tasks previously defined through the PLEXIL language. A C++ module has been developed to complete the simulation software.