Converged Artificial Intelligence and Robotics Technologies for Autonomous Industrial Operations and Smart Automation: A Structured Review of Architectures, Application Readiness, and Deployment Challenges

Rohit M. Kulkarni, Sneha P. Deshmukh, Amit R. Patil, Pooja S. Bhosale

Abstract


ABSTRACT
Industrial operations are moving from fixed, pre-programmed automation toward systems that perceive, reason and act with limited human intervention. This convergence of artificial intelligence (AI) and robotics, driven by deep perception, learning-based control, language-conditioned foundation models and digital-twin infrastructure, promises resilient and flexible production, yet its maturity, safety assurance and deployment barriers remain inconsistently documented. This review-based study synthesises 26 peer-reviewed, standards and industry sources published between 2007 and 2025 using a structured narrative protocol. A five-layer reference architecture is derived (sensing and actuation, edge perception, cognitive decision and learning, orchestration with digital twin, and enterprise and human interface), with safety assurance and data governance as cross-cutting concerns. Technology layers are compared, and six application domains are assessed through a four-criterion, five-point readiness rubric. Quality inspection (mean 4.1), predictive maintenance (4.0) and warehouse logistics (3.9) emerge as the most mature domains, whereas contact-rich assembly (3.3), welding and machining (3.1) and human-robot collaboration (2.9) lag, chiefly because decision autonomy (mean 3.3) trails perception reliability (mean 3.8). The findings indicate that supervised, perception-first autonomy is deployable today, while certified, learning-enabled decision autonomy is the principal open problem. Research priorities include sim-to-real validation, runtime safety monitoring and standardised industrial benchmarks.

KEYWORDS: Artificial intelligence, Industrial robotics, Autonomous operations, Smart automation, Digital twin, Industry 4.0, Reinforcement learning, Human-robot collaboration, Predictive maintenance, Cyber-physical systems

Full Text:

PDF 89-102