Multi-UAV Mapping and Target Finding in Large, Complex, Partially Observable Environments

Abstract

Coordinating multiple unmanned aerial vehicles (UAVs) for the purposes of target finding or surveying points of interest in large, complex, and partially observable environments remains an area of exploration. This work proposes a modeling approach and software framework for multi-UAV search and target finding within large, complex, and partially observable environments. Mapping and path-solving is carried out by an extended NanoMap library; the global planning problem is defined as a decentralized partially observable Markov decision process and solved using an online model-based solver, and the local control problem is defined as two separate partially observable Markov decision processes that are solved using deep reinforcement learning. Simulated testing demonstrates that the proposed framework enables multiple UAVs to search and target-find within large, complex, and partially observable environments.

Developing a Flying Explorer for Autonomous Digital Modelling in Wild Unknowns

“…In the explored method [21], the robot chooses the nearest, accessible and unvisited border with the largest grid size as its goal. The choice of the desired border is made by maximising the utility function (24), where f I i and f N i represent the grid size of the border area and the spatial distance between the robot and the border node, respectively. λ I and λ N are weighting parameters associated with these two terms.…”

“…As emphasised in [22], the consistent movement towards new frontiers allows the robot to extend its map into unexplored areas until the entire environment is covered. In unstructured environments, the existing boundary selection method may become inefficient [23] because robots tend to prioritise boundary points with high local information returns [24]. As a result, the robot may be attracted to these points, disregarding the continuity of environmental information in the current direction of exploration.…”