.png){kind=icon}
.png){kind=icon}
Return
Learning State-Space Models of Dynamic Systems from Arbitrary Data using Joint Embedding Predictive Architectures
.png){kind=icon}
PRE
AI
DOI：10.1016/j.ifacol.2025.10.190.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
With the advent of Joint Embedding Predictive Architectures (JEPAs), which appear to be more capable than reconstruction-based methods, this paper introduces a novel technique for creating world models using continuous-time dynamic systems from arbitrary observation data. The proposed method integrates sequence embeddings with neural ordinary differential equations (neural ODEs). It employs loss functions that enforce contractive em-beddings and Lipschitz constants in state transitions to construct a well-organized latent state space. The approachs effectiveness is demonstrated through the generation of structured latent state-space models for a simple pendulum system using only image data. This opens up a new technique for developing more general control algorithms and estimation techniques with broad applications in robotics. Copyright (c) 2025 The Authors. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
Keywords:
State-Space Models
Neural Networks
JEPA