Human-Centric Perception AI in Industrial Autonomous Systems – A Review

Abstract The rapid evolution of industrial autonomous systems has created a growing need for advanced perception capabilities that go beyond machine efficiency to prioritize human-centric values. Perception AI, encompassing multimodal sensing, contextual awareness, and adaptive decision-making, plays a pivotal role in enabling safe, collaborative, and trustworthy interactions between humans and machines in industrial environments. This study explores the integration of perception-driven intelligence into human-centric autonomous systems, emphasizing how advances in computer vision, including human detection and safety monitoring, pose, gesture, and activity recognition, emotion and intention, as well as Internet of Things, and sensor fusion models, enhance situational awareness and interpret human intention and behaviour. By framing perception AI as a bridge between technical autonomy and human-cantered design, this work highlights its transformative potential in analysing the development of the next generation of industrial automation that is not only efficient but also safe, interpretable, and socially aligned.