Learning Vision-Based Quadrotor Control in User Proximity

We consider a quadrotor equipped with a forward-facing camera, and an user freely moving in its proximity; we control the quadrotor in order to stay in front of the user, using only camera frames. To do so, we train a deep neural network to predict the drone controls given the camera image. Training data is acquired by running a simple hand-designed controller which relies on optical motion tracking data.


Published in:
369-369
Presented at:
2019 14th ACM/IEEE International Conference on Human-Robot Interaction (HRI), Daegu, Korea (South), March 11-14, 2019
Year:
Mar 25 2019
Keywords:
Other identifiers:
Laboratories:




 Record created 2019-10-31, last modified 2019-11-06


Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)