UAV-Based Camera Vibration Reduction for Detect and Avoid Tasks

Abstract: Unmanned aerial systems (UAS) have seen decreasing costs and increasing performance over several years, to the point that UAS technology is poised to make a dramatic positive impact on a wide range of commercial, industrial, and scientific endeavors. Although UAS technology has improved markedly from its state even five years ago, there are still significant technical challenges to be overcome if UAS is to become ubiquitous in civilian life. For obvious reasons, cameras have the potential to play a significant role in sense and avoid tasks, and it is common to see cameras deployed on drones. There are two common forms of disturbance that make cameras difficult to use on drones: first, the vibration of the motors creates a high frequency rolling shutter or “jello” effect. The resulting distortion in the image makes video analysis difficult or impossible. At the same time, lower frequency image distortions are common and due to multiple external disturbances: primarily aerodynamic effects from wind and the coupling of changes in attitude and position. Recent techniques developed for action cameras do a good job stabilizing video taken by a human. Other work has shown that an Inertial Measurement Unit (IMU) combined with a camera and calibrated using an Unscented Kalman Filter (UKF) can compensate for low-frequency distortion. At the same time, image processing can be used attenuate the high-frequency noise induced by the motors. We propose to combine these techniques to create an IMU-Augmented Jitter Reduction algorithm (IAJR) to stabilize video streams on unstable platforms such as UAVs. By approaching this problem in software, we expect to create a cost-effective method for drones to use camera data for detect and avoid tasks.

Details

  • Organization: NASA EPSCoR
  • Award #: NSHE-15-67
  • Amount: $36,512
  • Date: Nov. 18, 2015 - Aug. 31, 2016
  • PI: Dr. David Feil-Seifer
  • Co-PI: Dr. Richard Kelley

Supported Publications

  • Singandhupe, A., La, H., Feil-Seifer, D., Huang, P., Guo, L., & Li, M. Securing a UAV Using Individual Characteristics From an EEG Signal. In Proceedings of the IEEE Systems, Man, and Cybernetics Conference, page 2748-2753, Banff, Alberta, preprint, \url{https://arxiv.org/abs/1704.04574}. Oct 2017. preprint, \url{https://arxiv.org/abs/1704.04574}. ( details ) ( .pdf )