A Joint Data- and Model-Driven Approach Simulating the Behaviour of a Walkalong Glider

Abstract

Properly flying a walkalong glider is a challenging control problem, involving internal and external influences. We approach this problem from both a model-driven and data-driven viewpoint. We construct a mathematical model, based on well-established theory, which is used to derive both optimal paddle angle and distance between the paddle and the glider for steady level flight. Moreover, we derive parameters of the glider during uniform circular flight, as well as their relation to movement of the paddle from our model. As the glider is in practice very sensitive to airflow disturbances, we develope a data-driven model that tracks the status of the glider using video data, recorded by a camera mounted on the paddle. We use image processing and deep learning to automatically detect glider position with respect to the paddle and its yaw. Both approaches could be jointly used for feedback controlling an automatically operating walkalong glider.

Authors 7

1. Dieuwertje Alblas

   University of Twente

   Affiliation as printed

   University of Twente

2. Mark van den Bosch Aachen

   Leiden University

   Affiliation as printed

   Leiden University

3. Vivi Rottschäfer Aachen

   Leiden University

   Affiliation as printed

   Leiden University

4. Felix L. Schwenninger

   University of Twente

   Affiliation as printed

   University of Twente

5. Len Spek

   University of Twente

   Affiliation as printed

   University of Twente

6. Lotte Weedage

   University of Twente

   Affiliation as printed

   University of Twente

7. Sjanne Zeijlemaker

   Eindhoven University of Technology

   Affiliation as printed

   Eindhoven University of Technology