Human-aware Contingent Planning.

Contingent planning models a robot that must achieve a goal in a partially observable environment with non-deterministic actions. A solution for this problem is generated by searching in the space of belief states, where a belief state is a set of possible world states. However, if there is an unavoidable dead-end state, the robot will fail to accomplish his task. In this work, rather than limiting a contingent planning task to the agent's actions and observations, we model a planning agent that is able to proactively resort to humans for help in order to complete tasks that would be unsolvable otherwise. Our aim is to develop a symbiotic autonomous agent, that is, an agent that, proactively and autonomously, asks for human help when needed. We formalize this problem and propose an extension of a translation technique to convert the contingent planning problem with human help into a non-deterministic fully observable planning problem that can be solved by an off-the-shelf efficient FOND planner.