地図に依存しない統合的な点検・探索ミッションのためのフレームワーク
Towards a Reduced Dependency Framework for Autonomous Unified Inspect-Explore Missions
未知環境で小型ドローンが限られたセンサのみを用い、地図に頼らずに構造物を探索・近接点検して3次元形状を取得する統合的な計画手法を提案し、屋内実験で検証した。
著者: Vignesh Kottayam Viswanathan, Sumeet Gajanan Satpute, Ali-akbar Agha-mohammadi, George Nikolakopoulos
分類: cs.RO
原文アブストラクト
The task of establishing and maintaining situational awareness in an unknown environment is a critical step to fulfil in a mission related to the field of rescue robotics. Predominantly, the problem of visual inspection of urban structures is dealt with view-planning being addressed by map-based approaches. In this article, we propose a novel approach towards effective use of Micro Aerial Vehicles (MAVs) for obtaining a 3-D shape of an unknown structure of objects utilizing a map-independent planning framework. The problem is undertaken via a bifurcated approach to address the task of executing a closer inspection of detected structures with a wider exploration strategy to identify and locate nearby structures, while being equipped with limited sensing capability. The proposed framework is evaluated experimentally in a controlled indoor environment in presence of a mock-up environment validating the efficacy of the proposed inspect-explore policy.
