A Modeling Strategy Using Bayesian Optimization with POMDP for Exploration and Informative Path Planning of UAVs in Monitoring of Forest

Unmanned Aerial Vehicles (UAVs) have been used for several applications in monitoring complex and unknown environments. The challenge is to plan missions for the UAV in situations where the vehicle needs to visit and explore an area and analyze it in real-time to define the route to be followed. This visit occurs with the search area's maximization from the definition of the trajectories, making it possible to collect information to acquire knowledge about the environment and provide a map. This type of problem is known as Informative Path Planning (IPP) and Autonomous Exploration (AE). In this context, Bayesian Optimization (BO) has been adopted. Moreover, there is a need to define the planning for decision-making based on information about the environment considering specific restrictions. Partially Observable Markov Decision Processes (POMDP) can be used to define the planner responsible for the decision-making. Therefore, considering these methods and the strategy of exploring the environment in a continuous 3D space, this paper proposes a development of a modeling strategy that is exploration and informative path planning using Sequential Bayesian Optimization with POMDP in forest monitoring with Canopy gap. A sequential decision-maker was developed under uncertainty based on the Sequential Bayesian Optimization approach with Partially Observable Markov Decision Process (BO-POMDP).