Towards Robust Mission Execution via Temporal and Contingent Planning.

In this work, we present a general approach to task planning based on the combination of temporal planning, contingent planning and run-time sensing. The strategy provides a solution for generating plans that can adapt during mission execution by reasoning about the data acquired by the sensory system. The approach detects actions that can change the initial plan and evaluates possible outcomes. We demonstrate the effectiveness of our approach on two different experiments in a maritime environment where the robots have to inspect the state of a valve and execute actions based on the online sensor information.