A Computationally Efficient UAV Exploration Strategy for Uncertain Complex Environments

The paper proposes a frontier based strategy for exploration. Unlike traditional frontier based approaches, a computationally efficient approach is proposed, which does not require to process all the frontiers in a map frequently, and is suitable for on-board computing. To achieve this, the global set of frontiers is divided into three subsets. Each subset is associated with a distinct planning strategy, which allows exploration in a computationally efficient way. Complexity analysis conveys the superiority of the proposed approach in terms of computational load. Performance of the proposed approach is validated in simulation by employing a quadrotor in large-scale and complex problem setups, e.g., powerplant, bridge structure, and typical office like scenario. Simulation results confirm the efficacy of the proposed approach over state-of-the-art (SoA) algorithms.