Adaptive Path Planning for UAV-based Multi-Resolution Semantic Segmentation

In this paper, we address the problem of adaptive path planning for accurate\nsemantic segmentation of terrain using unmanned aerial vehicles (UAVs). The\nusage of UAVs for terrain monitoring and remote sensing is rapidly gaining\nmomentum due to their high mobility, low cost, and flexible deployment.\nHowever, a key challenge is planning missions to maximize the value of acquired\ndata in large environments given flight time limitations. To address this, we\npropose an online planning algorithm which adapts the UAV paths to obtain\nhigh-resolution semantic segmentations necessary in areas on the terrain with\nfine details as they are detected in incoming images. This enables us to\nperform close inspections at low altitudes only where required, without wasting\nenergy on exhaustive mapping at maximum resolution. A key feature of our\napproach is a new accuracy model for deep learning-based architectures that\ncaptures the relationship between UAV altitude and semantic segmentation\naccuracy. We evaluate our approach on the application of crop/weed segmentation\nin precision agriculture using real-world field data.\n