Network-based Active Inference and its Application in Robotics

ICLR 2025 Conference Submission12637 Authors

Keywords: Active Inference (AIF), Free Energy Principle (FEP), Robotics, Trajectory generation, Random dynamical systems, Random attractor dynamics, Non-Equilibrium Steady State (NESS), Adaptive control, Industrial automation, Computational efficiency, Cost-efficient solutions

TL;DR: Network-based Active Inference (NetAIF) is a novel framework that uses random attractor dynamics and the Free Energy Principle to enable real-time adaptive robotics in unstructured environments.

Abstract: This paper introduces Network-based Active Inference (NetAIF), a novel robotic framework that enables real-time learning and adaptability in dynamic, unstructured environments. NetAIF leverages random attractor dynamics and the Free Energy Principle (FEP) to simplify trajectory generation through network-topology-driven attractors that induce controlled instabilities and probabilistic sampling cycles. This approach allows robots to efficiently adapt to changing conditions without requiring extensive pre-training or pre-calculated trajectories. By integrating learning and control mechanisms within a compact model architecture, NetAIF facilitates seamless task execution, such as target tracking and valve manipulation. Extensive simulations and real-world experiments demonstrate NetAIF's capability to perform rapid and precise real-time adjustments, highlighting its suitability for applications requiring high adaptability and efficient control, such as robotics tasks in the energy and manufacturing sectors.

Primary Area: probabilistic methods (Bayesian methods, variational inference, sampling, UQ, etc.)

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 12637