Uncertainty-Aware Multi-Robot Flocking via Learned State Estimation and Control Barrier Functions
Mattia Catellani, Lorenzo Sabattini
Abstract
Information exchange is crucial for optimal coordination of robots, but a link may not always be available among agents to share data. For this reason, this paper presents a decentralized solution for flocking control, leveraging state and uncertainty estimation of undetected robots. A neural network is trained to mimic a state estimator, also providing information about the uncertainty of the estimate. This uncertainty is used to weigh the contribution of the estimate in taking actions for coordination. Using Control Barrier Functions and Control Lyapunov Functions, we define an optimization problem to find an optimal control input to reproduce collective motion observed in nature. We evaluate both the learned estimator and the control strategy with extensive simulations.
BibTeX
@inproceedings{iros2025_uncertaintyaware,
title = {Uncertainty-Aware Multi-Robot Flocking via Learned State Estimation and Control Barrier Functions},
author = {Mattia Catellani and Lorenzo Sabattini},
booktitle = {IROS 2025},
year = {2025}
}