This repository contains the software tool discussed in the paper:
Passive Fault-Tolerant Augmented Neural Lyapunov Control: a method to synthesise control functions for marine vehicles affected by actuators faults
The work can be read open-access here.
pFT-ANLC is a software tool to automatically synthesise:
The code is based on a loop between a Learner and a Falsifier. Starting from a finite set of state-space samples, the Learner trains two Artificial Neural Networks (ANNs), one representing a control law and the other a CLF.
In parallel, the Falsifier is tasked with verifying whether the candidate CLF satisfies the theoretical Lyapunov conditions within the dense domain over the Reals.
If the theoretical Lyapunov conditions are satisifed, the learning is halted and the resulting control law and CLF are returned. If the conditions are not satisfied, the Falsifier returns a set of points (denoted as counterexample) where the Lyapunov conditions are violated. These points are added to the dataset and the learning process is further iterated.
The learning system attempts to simultaneously stabilise a set of dynamics, ecompassing the nominal dynamics (fault-free) and the faulty modes.
A schematic of the learning architecture is hereby illustrated:
The pFT-ANLC tool features:
Instructions on installation are available within the file.
To synthesise control laws and CLFs for your own dynamics, a step-by-step example is reported in the file.
Hereby an example of how a CLF and a corresponding Lie derivative function are updated over successive training iterations.
Note the yellow patches of the Lie derivative gradually disappearing as the training proceeds. The training halts once, at the same time, the CLF is certified to be positive definite and the Lie derivative is certified to be negative definite.
CLF evolution | Lie derivative function evolution |
---|---|
The library architecture is composed of three main modules:
The code currently supports:
utilities/Function/AddLieViolationsOrder4_v4
. The authors can be contacted for feedback, requests of clarifications or requests of support at:
grande.rdev@gmail.com
The article can be read open-access here.
This work can be cited with the following BibTeX entry:
@article{grande2024passive,
title={Passive Fault-Tolerant Augmented Neural Lyapunov Control: A method to synthesise control functions for marine vehicles affected by actuators faults},
author={Grande, Davide and Peruffo, Andrea and Salavasidis, Georgios and Anderlini, Enrico and Fenucci, Davide and Phillips, Alexander B and Kosmatopoulos, Elias B and Thomas, Giles},
journal={Control Engineering Practice},
volume={148},
pages={105935},
year={2024},
publisher={Elsevier}
}