We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

cs

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Learn Proportional Derivative Controllable Latent Space from Pixels

Abstract: Recent advances in latent space dynamics model from pixels show promising progress in vision-based model predictive control (MPC). However, executing MPC in real time can be challenging due to its intensive computational cost in each timestep. We propose to introduce additional learning objectives to enforce that the learned latent space is proportional derivative controllable. In execution time, the simple PD-controller can be applied directly to the latent space encoded from pixels, to produce simple and effective control to systems with visual observations. We show that our method outperforms baseline methods to produce robust goal reaching and trajectory tracking in various environments.
Subjects: Machine Learning (cs.LG)
Cite as: arXiv:2110.08239 [cs.LG]
  (or arXiv:2110.08239v1 [cs.LG] for this version)

Submission history

From: Weiyao Wang [view email]
[v1] Fri, 15 Oct 2021 17:47:07 GMT (5508kb,D)

Link back to: arXiv, form interface, contact.