Skip to main content
Journal cover image

State Primitive Learning to Overcome Catastrophic Forgetting in Robotics

Publication ,  Journal Article
Xiong, F; Liu, Z; Huang, K; Yang, X; Qiao, H
Published in: Cognitive Computation
March 1, 2021

People can learn continuously a wide range of tasks without catastrophic forgetting. To mimic this functioning of continual learning, current methods mainly focus on studying a one-step supervised learning problem, e.g., image classification. They aim to retain the performance of previous image classification results when neural networks are sequentially trained on new images. In this paper, we concentrate on solving multi-step robotic tasks sequentially with the proposed architecture called state primitive learning. By projecting the original state space into a low-dimensional representation, meaningful state primitives can be generated to describe tasks. Under two kinds of different constraints on the generation of state primitives, control signals corresponding to different robotic tasks can be separately addressed only with an efficient linear regression. Experiments on several robotic manipulation tasks demonstrate the new method efficacy to learn control signals under the scenario of continual learning, delivering substantially improved performance over the other comparison methods.

Duke Scholars

Published In

Cognitive Computation

DOI

EISSN

1866-9964

ISSN

1866-9956

Publication Date

March 1, 2021

Volume

13

Issue

2

Start / End Page

394 / 402

Related Subject Headings

  • 1702 Cognitive Sciences
  • 1109 Neurosciences
  • 0801 Artificial Intelligence and Image Processing
 

Citation

APA
Chicago
ICMJE
MLA
NLM
Xiong, F., Liu, Z., Huang, K., Yang, X., & Qiao, H. (2021). State Primitive Learning to Overcome Catastrophic Forgetting in Robotics. Cognitive Computation, 13(2), 394–402. https://doi.org/10.1007/s12559-020-09784-8
Xiong, F., Z. Liu, K. Huang, X. Yang, and H. Qiao. “State Primitive Learning to Overcome Catastrophic Forgetting in Robotics.” Cognitive Computation 13, no. 2 (March 1, 2021): 394–402. https://doi.org/10.1007/s12559-020-09784-8.
Xiong F, Liu Z, Huang K, Yang X, Qiao H. State Primitive Learning to Overcome Catastrophic Forgetting in Robotics. Cognitive Computation. 2021 Mar 1;13(2):394–402.
Xiong, F., et al. “State Primitive Learning to Overcome Catastrophic Forgetting in Robotics.” Cognitive Computation, vol. 13, no. 2, Mar. 2021, pp. 394–402. Scopus, doi:10.1007/s12559-020-09784-8.
Xiong F, Liu Z, Huang K, Yang X, Qiao H. State Primitive Learning to Overcome Catastrophic Forgetting in Robotics. Cognitive Computation. 2021 Mar 1;13(2):394–402.
Journal cover image

Published In

Cognitive Computation

DOI

EISSN

1866-9964

ISSN

1866-9956

Publication Date

March 1, 2021

Volume

13

Issue

2

Start / End Page

394 / 402

Related Subject Headings

  • 1702 Cognitive Sciences
  • 1109 Neurosciences
  • 0801 Artificial Intelligence and Image Processing