Planning is an essential element of effective autonomous robot operation. Planning improves robot operation performance by providing an optimized action schedule based on optimization criteria such as time, energy or local restrictions. One step further is the usage of multi-robot teams which allows to solve more complex problems more effectively. But multi-robot planning is particulary challenging due to the increased computational cost in the planning process, communication requirements during the execution process but also the risk of mission failures due to failures of one individual robot.
In my Master's Thesis I aim to apply proactive planning approaches to reconfigurable multi-robot missions. By applying (probabilistic) metrics such as reliability, mean task duration or energy consumption I want to find potential weaknesses in the proposed mission and improve the overall mission execution performance by allowing the system to react proactively with the help of prepared reactions. For this I want to implement a mission evaluation software which simulates the mission execution process, analyzes overall mission performance and identifies potential risks of failures. These can then be utilized to generate strategies for proactive failure handling such as generation of fallback plans or plan repair.