WP2: Enabling Open-Ended Safe Self-Adaptation

Objective

Develop the necessary tools for enabling open-ended self-adaptation within the MAPE-K loop, covering monitoring, analysis, learning, validation, and knowledge modelling. Supports objectives O1 and O3.

Lead: AUTH · Effort: 136 person-months · Duration: M1–M36

Research Tasks

T2.1: Autonomous System Monitoring (M1–M36) — Development of an uncertainty estimator and anomaly detection mechanisms for continuous monitoring of autonomous systems.

T2.2: Active Environmental State Analysis (M1–M36) — Active perception techniques and an anomaly detection taxonomy for analysing the environmental state surrounding the autonomous system.

T2.3: Autonomous System Learning and Adaptation (M1–M36) — Deep learning approaches for handling structural and environmental changes, enabling the system to learn and adapt autonomously.

T2.4: Safety and Performance Validation (M1–M36) — Functional and safety requirements specification, evidence-based evaluation of self-adaptation actions.

T2.5: Knowledge Modeling and Data/AI Standards (M1–M36) — Ensuring data consistency and compliance with ONNX standards for interoperable knowledge representation.

Deliverables

Monitoring tools, analysis frameworks, learning algorithms, validation methods, and knowledge models.