Ulmen, Jonas (RPTU Kaiserslautern-Landau), Sundaram, Ganesh (RPTU University Kaiserslautern-Landau, Germany), Görges, Daniel (University of Kaiserslautern)

Learning State-Space Models of Dynamic Systems from Arbitrary Data Using Joint Embedding Predictive Architectures

Scheduled for presentation during the Regular Session "AI-based Robot Control I" (WeAT2), Wednesday, July 16, 2025, 11:00−11:20, Room 106

Joint 10th IFAC Symposium on Mechatronic Systems and 14th Symposium on Robotics, July 15-18, 2025, Paris, France

This information is tentative and subject to change. Compiled on July 16, 2025

Abstract

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 embeddings and Lipschitz constants in state transitions to construct a well-organized latent state space. The approach's 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.