Print Email Facebook Twitter Vision-Based Reinforcement Learning for the guidance of an AR Drone 2 Title Vision-Based Reinforcement Learning for the guidance of an AR Drone 2 Author Siddiquee, Manan (TU Delft Aerospace Engineering) Contributor van Kampen, E. (mentor) Junell, J. (mentor) Degree granting institution Delft University of Technology Programme Aerospace Engineering Date 2018-04-17 Abstract Reinforcement Learning (RL) has been applied to teach quadcopters guidance tasks. Most applications rely on position information from an absolute referencesystem such as Global Positioning System (GPS). The dependence on "absoluteposition" information is a general limitation in the autonomous flight of Unmanned Aerial Vehicles (UAVs). Environments that have weak or no GPS signals are difficult to traverse for them. Instead of using absolute position, it ispossible to sense the environment and the information contained within it inorder to come up with a "relative" description of the UAV's position. Thispaper presents the design of a RL agent with relative vision-based states and rewards for the teaching of a guidance task to a quadcopter. The agent is taught the task of turning towards a red marker and approaching it in simulation and in flight tests. A more complex task of travelling between a blue and a red marker is trained in simulation. This work shows that relative vision-based states and rewards can be used with RL to teach quadcopters simple guidance tasks. The performance of the trained agent is inconsistent in simulation and flight test due to the inherent partialobservability in the relative description of the state. Subject Reinforcement LearningComputer VisionGuidanceNavigationQ-LearningDroneflight testing To reference this document use: http://resolver.tudelft.nl/uuid:334e0430-17b8-4366-b8b6-0a5f8fb0367a Part of collection Student theses Document type master thesis Rights © 2018 Manan Siddiquee Files PDF msiddiquee_thesisreport_F ... il2018.pdf 6.17 MB Close viewer /islandora/object/uuid:334e0430-17b8-4366-b8b6-0a5f8fb0367a/datastream/OBJ/view