A deep neural network approach to fusing vision and heteroscedastic motion estimates for low-SWaP robotic applications.
Jared ShamwellWilliam D. NothwangDonald PerlisPublished in: MFI (2017)
Keyphrases
- neural network
- motion estimates
- structure from motion
- motion estimation
- computer vision
- optical flow
- error analysis
- motion model
- estimation accuracy
- image sequences
- image noise
- bundle adjustment
- camera motion
- prediction error
- optical flow estimation
- vision system
- motion vectors
- feature tracking
- prediction accuracy
- super resolution
- ego motion
- three dimensional
- machine learning