A structured taxonomy of autonomous manufacturing systems: Capabilities, definitions, and enabling technologies

Smart manufacturing has introduced adaptive, data-driven systems to the shop floor and beyond. However, current deployments operate near their technological limits, lacking the contextual adaptability, resilience, and independent decision-making required to manage unplanned disruptions. Autonomous manufacturing systems have emerged as a promising solution to overcome these deficiencies. These systems fulfill industry’s growing need for proactive rather than reactive approaches, balancing efficiency, safety, and agility. However, the field remains fragmented by inconsistent definitions, inadequate classification frameworks, and unclear capability-technology mappings. Terms like adaptive, autonomous, intelligent, smart, and cognitive systems are used interchangeably, creating confusion that hinders broader development and adoption. Critically, manufacturing lacks domain-specific autonomy taxonomies comparable to automotive industry standards, preventing systematic capability specification and solution comparison. Existing taxonomies frequently conflate service robotics and military systems with manufacturing applications, which demand precision, repeatability, and interoperability. This paper presents a manufacturing-specific autonomy taxonomy developed through analysis of cross-domain industrial taxonomies and maturity frameworks. The taxonomy establishes a six-tier classification: Cyber-Physical, Intelligent, Smart (semi and fully mature tiers), Autonomous, and Cognitive manufacturing systems. Classic automated manufacturing systems and adaptive manufacturing systems are pre-autonomy baselines below the entry tier and are excluded from the six-tier count. Each tier is defined by distinct capabilities, decision-making authority, learning mechanisms, enabling technologies, and human-machine collaboration requirements. First, this work provides standardized terminology and capability-technology mappings, functioning as an analytical tool rather than an absolute classification that divides the world into rigid classes. Second, it aims to serve as a structured, practical reference to guide practitioners and stakeholders in defining system objectives, identifying enabling technologies, and aligning operational requirements with appropriate autonomy levels. Ultimately, the proposed taxonomy establishes a common benchmark for system development and evaluation toward achieving autonomous and cognitive production.