1 min read 187 words

RELIABLE

Advances in control design methodologies for safety-critical systems applied to robotics

control systems robotics safety-critical systems machine learning autonomous systems

Table of Contents

Overview

RELIABLE aims to develop control system methodological tools and algorithms that explicitly integrate safety constraints in their formulation and are provably (mathematically) safe certified. A particular goal is to combine data-driven approaches and machine learning techniques with recent optimization-based control techniques capable of enforcing invariance in the context of Control Barrier Functions (CBFs) and Control Lyapunov Functions (CLFs) in the presence of challenging restrictions and uncertainties.

The project also addresses the transition from single systems to large-scale safety-critical networked systems involving multiple agents operating autonomously over networks in dynamic environments, where additional challenges arise due to the presence of communication networks.

Objectives

Develop control methodologies with formally certified safety guarantees
Combine data-driven and machine learning approaches with optimization-based control
Apply Control Barrier Functions (CBFs) and Control Lyapunov Functions (CLFs) under uncertainty
Scale methodologies from single systems to networked multi-agent systems
Address communication network challenges in distributed autonomous systems

Case Studies

Robotic vehicles: Space, aerial, and underwater scenarios for remote monitoring and exploration applications
Mobile robotics in Industry 4.0: Perception algorithms, reactive planning, navigation and control systems enabling autonomous operation in unstructured environments with safe human-robot collaboration