Unified Modelling of Intelligent Robotic Systems: Applications of COH/GISMOL in Automation Engineering

Robotics and automation increasingly demand intelligent systems that integrate perception, cognition, and action while assuring safety and compliance in dynamic environments. Existing paradigms—behavior-based control, layered architectures, middleware frameworks, and learning-based controllers—solve parts of the problem but rarely offer a cohesive way to model structure, learning, and constraints together. This paper introduces Constrained Object Hierarchies (COH) and its Python toolkit GISMOL (General Intelligent System Modelling Language) as a unified approach to designing and implementing intelligent systems. COH formalizes intelligent systems via a 9‑tuple representation—components, attributes, methods, neural components, embeddings, identity constraints, trigger constraints, goal constraints, and constraint daemons —thereby separating what a system is from what it does and what must never be violated. Neuroscience‑inspired hierarchical processing motivates this decomposition and the separation of learned behaviors from innate constraints. We demonstrate COH/GISMOL on five representative cases in robotics and automation: warehouse AMR, collaborative assembly, autonomous harvesting, predictive maintenance, and multi‑robot search‑and‑rescue. Across these cases, COH/GISMOL delivers: (i) systematic modelling, (ii) integrated learning with safety‑first constraint enforcement, and (iii) maintainable hierarchies that support verification and runtime monitoring consistent with emerging standards and best practices.