Thu, Jun 20
|Rockville
Drone Group Learning Series: The Flapper Drone and Estimating Distance in Flaps
Please RSVP here: https://www.meetup.com/dc-area-drone-user-group/events/301445354 Levi will be presenting on Flapper Drones (including a demonstration). Afterwards, Mike will have his drone racing controller and we'll set up the simulator on the big TV for anyone interested to try out.
Time & Location
Jun 20, 2024, 6:00 PM – 8:00 PM
Rockville, 33 Maryland Ave, Rockville, MD 20850, USA
About the event
Please RSVP here: https://www.meetup.com/dc-area-drone-user-group/events/301445354
Levi will be presenting on Flapper Drones (including a demonstration). Afterwards, Mike will have his drone racing controller and we'll set up the simulator on the big TV for anyone interested to try out.
We'll get started right after 6pm so please try to arrive on time!
What is a Flapper Drone? Let’s say at your desk, looking at the objects around you on the table. You have the feeling that you know where they are, but you don’t know their distance from your eye in meters or feet. Instead, you know where the objects are because you can immediately move your arm and touch anything in front of you. This basic principle is how we, and all animals, encode distances as the actions we must take to reach something. My dissertation studies this intuition and proposes a set of equations that allow testing the theory on robots such as the Flapper Drone.
The Flapper Drone is a commercially available ornithopter whose current applications are limited. Unlike a quadrotor, it cannot fly itself using onboard cameras. It simply shakes too much for traditional computer vision algorithms to track objects in the video feed. However, by adding sensors to the wings that measure their motion, we can predict the shaking using the actions (flaps) that occur and partially stabilize the image. This algorithm naturally estimates distances to nearby objects in the units of the motion flapping, or the motoric units, similar to how humans sometimes estimate distance in the steps it would take to traverse it. Through this bio-inspired approach, we aim to produce an algorithm that allows the Flapper to fly itself using the stabilized image.
In the talk, we will introduce the Flapper, its sensors and capabilities, the recent history of flapping robots, and do a live-demo. Then we will discuss the challenges the Flapper's shaking introduces and how the principle of action based distance (or visuomotor embodiment) is helping us solve them. We will conclude by briefly discussing a book aimed at technically inclined makers who wish to make their own ornithopters called "The Delfly".