Published In

Frontiers in Artificial Intelligence and Applications

Document Type

Article

Publication Date

9-21-2026

Subjects

Autonomous vehicles, Fault recovery, Gaussian Process, PID control

Abstract

Unmanned autonomous vehicles (UAVs) must often operate reliably in remote, hostile environments for extended periods. Fault tolerance is essential, enabling continued operation despite system faults. Fault recovery is challenging due to unknown fault characteristics and no possibility of on-site human intervention. Replacing faulty hardware with operating spares is an effective recovery method, but impractical because UAVs have limited space. Reconfiguration may be the only practical option. This paper discusses correcting faulty behavior in robot arm DC servos, guided by a PID controller, by modifying the controller’s gain values. Although simple in principle, finding new gain values is difficult and time-consuming, which increases the risk of further robot arm damage. We use a Gaussian Process to probabilistically model how particular gain values affect robot arm behavior. The described method quickly and effectively finds new gain values. Two fault recovery examples are presented. This is, to the best of our knowledge, the first time a Gaussian Process has been used in a fault tolerant application.

Rights

Copyright (c) 2026 The Authors Creative Commons License This work is licensed under a Creative Commons Attribution 4.0 International License.

DOI

10.3233/FAIA260933

Persistent Identifier

https://archives.pdx.edu/ds/psu/45325

Publisher

IOS Press

Download

Share

COinS