マイクロUAVのための単眼・ステレオ手がかりによる着陸地点評価
Monocular and Stereo Cues for Landing Zone Evaluation for Micro UAVs
ステレオ画像から斜面を、単眼画像からホモグラフィ誤差を算出し組み合わせることで、剛体・非剛体の動的表面に対応した高速・高精度なUAV着陸地点評価手法を提案した。
著者: Rohit Garg, Shichao Yang, Sebastian Scherer
分類: cs.RO
原文アブストラクト
Autonomous and safe landing is important for unmanned aerial vehicles. We present a monocular and stereo image based method for fast and accurate landing zone evaluation for UAVs in various scenarios. Many existing methods rely on Lidar or depth sensor to provide accurate and dense surface reconstruction. We utilize stereo images to evaluate the slope and monocular images to compute homography error. By combining them together, our approach works for both rigid and non-rigid dynamic surfaces. Experiments on many outdoor scenes such as water, grass and roofs, demonstrate the robustness and effectiveness of our approach.
