PVEs: Position-Velocity Encoders for Unsupervised Learning of Structured State Representations

@article{Jonschkowski2017PVEsPE,
  title={PVEs: Position-Velocity Encoders for Unsupervised Learning of Structured State Representations},
  author={Rico Jonschkowski and Roland Hafner and Jonathan Scholz and Martin A. Riedmiller},
  journal={CoRR},
  year={2017},
  volume={abs/1705.09805}
}
We propose position-velocity encoders (PVEs) which learn—without supervision—to encode images to positions and velocities of task-relevant objects. PVEs encode a single image into a low-dimensional position state and compute the velocity state from finite differences in position. In contrast to autoencoders, position-velocity encoders are not trained by image reconstruction, but by making the position-velocity representation consistent with priors about interacting with the physical world. We… CONTINUE READING
Related Discussions
This paper has been referenced on Twitter 22 times. VIEW TWEETS

Citations

Publications citing this paper.
Showing 1-10 of 16 extracted citations

SE3-Pose-Nets: Structured Deep Dynamics Models for Visuomotor Control

2018 IEEE International Conference on Robotics and Automation (ICRA) • 2018
View 2 Excerpts

References

Publications referenced by this paper.
Showing 1-10 of 25 references

SE3-nets: Learning rigid body motion using deep neural networks

2017 IEEE International Conference on Robotics and Automation (ICRA) • 2017
View 1 Excerpt

End-to-end learnable histogram filters. In Workshop on Deep Learning for Action and Interaction at NIPS, December 2016

Rico Jonschkowski, Oliver Brock
2016
View 1 Excerpt

FaceNet: A unified embedding for face recognition and clustering

2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) • 2015
View 1 Excerpt

Similar Papers

Loading similar papers…