Autonomous DelFly

A lightweight, yet fully autonomous insect-inspired flapping wing robot controlled by an onboard autopilot computer and a stereo vision system. Its capabilities were demonstrated in a multiroom exploration task.

In this project, carried out together with Kirk Scheper, we aimed at extending the autonomous capabilities of the DelFly Explorer of MAVLab, TU Delft. The robot was upgraded with a Lisa/S autopilot, allowing to run an open-source Paparazzi UAV autopilot software. The mission chosen to demonstrate the robot’s capabilities consisted of 2 tasks: room exploration and door traversal. In the exploration part, the vehicle was flying in a room, while following its walls (using the droplet strategy) and searching for a door (using a contrast-based “snake gate” detector). Once a door was found, the vehicle navigated towards the door and traversed into the next room, where it returned back into to the exploration mode. The major challenges of this project included the development of computationally efficient algorithms for computer vision and navigation, as well as introducing flight testing procedures that minimize crashes of the relatively delicate robot. The research is summarized in the following video, made and narrated by Kirk:

For more details, check out our conference paper.

Related publications:

  • K. Y. W. Scheper, M. Karásek, C. {De Wagter}, B. D. W. Remes, and G. C. H. E. de Croon, “First autonomous multi-room exploration with an insect-inspired flapping wing vehicle,” in 2018 International Conference on Robotics and Automation, Brisbane, Australia, 2018, p. 7.
    address = {Brisbane, Australia},
    author = {Scheper, Kirk Y W and Kar{\'{a}}sek, Mat{\v{e}}j and {De Wagter}, Christophe and Remes, Bart D. W. and de Croon, Guido C. H. E.},
    booktitle = {2018 International Conference on Robotics and Automation},
    pages = {7},
    title = {{First autonomous multi-room exploration with an insect-inspired flapping wing vehicle}},
    year = {2018}