Towards a Probabilistic Based Autonomous UAV Mission Planning for Planetary Exploration

The level of autonomy on Unmanned Aerial Vehicles (UAV) to conduct different missions has remarkably increased over the past ten years. However, a higher level of autonomy is needed when UAVs need to deal with uncertainty derived from lack of Global Position Systems, complex exploration missions in remote partially observable environments and sensor failures. Probabilistic techniques such as the Partially Observable Markov Decision Process (POMDP) have been used to deal with uncertainty in the environment and observations from onboard and payload sensors in robotic applications. In this work, we approach the problem of UAV autonomous mission planning and execution for planetary exploration from the perspective and advantages that probabilistic based planning techniques can bring to a remote exploration mission using UAVs. The POMDP mission planning formulation was implemented and tested in a simulated environment with exploration targets and terrain. Furthermore, future and current work to improve the simulation environment and the autonomous mission planning capabilities for planetary exploration using UAVs is discussed.