AI-Survey for Self-Flying Vehicles: Exploring the Challenges of Deep LearningDownload PDF

Oct 12, 2018 (edited Nov 22, 2018)NIPS 2018 Workshop MLITS SubmissionReaders: Everyone
  • Keywords: Autonomous Flights, 3D Object Detection, Semantic Segmentation, End-2-End-Flying
  • TL;DR: Compressed AI Survey for Autonomous Flights
  • Abstract: Everyone is talking about intuitive and automated transportation. An important and very challenging part of this research field are autonomous unmanned aerial vehicles (UAV) such as automated air taxis with a vertical take-off and landing (VTOL) capability. On one hand autonomous VTOLs will redesign our personal understanding of urban mobility, on the other hand automated UAVs will drastically change any kind of delivery or transportation services and much more. However, when studying computer vision and machine learning problems for UAVs or VTOLs it becomes increasingly difficult to stay up-to-date. We provide a survey for the topic of automated flights focusing on challenging Deep Learning problems with a state-of-the-art overview. We give an outline of possible sensor set-ups and AI based pipelines with leading results on established data sets. Finally, we point out currently missing investigations.
2 Replies