Decision Making with STPA through Markov Decision Process, a Theoretic Framework for Safe Human-Robot Collaboration

During the last decades, collaborative robots capable of operating out their cages are widely used in industry to assist humans mundane and harsh manufacturing tasks. Although such inherently safe by design, they commonly accompanied external sensors other cyber-physical systems, facilitate close cooperation with humans, which frequently render ecosystem unsafe prone hazards. We introduce a method that capitalizes on partially observable Markov decision processes (POMDP) amalgamate nominal actions system along control posed System Theoretic Process Analysis (STPA). A decision-making mechanism constantly prompts into safer state is realized providing situation awareness about safety levels associating specific groups selected actions. POMDP compensates partial observability uncertainty current environment creates screening policies tend make decisions balance from states real time during operational phase. The theoretical framework assessed simulated human–robot scenario proved identifying loss success scenarios.