Abstract:
Planning is a sub discipline of artificial intelligence that aims at generating plans so that its execution leads to a desirable state from some initial state. When uncertainty is involved in a planning domain, the planning problem is no longer limited to finding a sequence of actions, but to find a plan, also composed of actions, that might include conditional, iterative and/or recursive_ constructs, i.e. the planning problem now also includes program synthesis. Generating programs automatically is a challenging task. Even for simple planning domains, like the blocks .world domain involving recursion and if-then-else, the problem of generating programs needs human intervention. In the dissertation, a modified blocks world domain is considered and planners are developed for the domain. Also, some results for the plans generated by the planners are presented.
Try actions are the trials of physical actions. As another part of the dissertation, the concept of try actions has been extended to joint trials, trials involving more than one agent. Other concepts relating to trials, repeated trials and learning about the capabilities of agents have also been introduced.
