Cybersecurity Metrics for Human-Robot Collaborative Automotive Manufacturing

Abstract

A human-robot collaborative system was developed where a human co-worker and a collaborative robot collaborated to perform assembly tasks. We assumed such assembly tasks as the representative assembly tasks in the automotive manufacturing industry. We assumed that the collaborative assembly system would be a part of an industrial control system (ICS) in actual industrial environment. We recruited 20 human subjects to perform the assembly tasks in collaboration with the robot. We trained the subjects to enable them to perform the collaborative assembly tasks, and evaluate cybersecurity conditions (status) of the collaborative system. Each subject performed the collaborative assembly separately with the robot. At the end of the assembly task, each subject was asked to brainstorm, write down the cybersecurity assessment criteria as best as he/she could perceive or realize through his/her experiences with the collaborative system, and rate the cybersecurity status of the collaborative system for each of the criteria (i.e., how capable the collaborative system was in fulfilling the cybersecurity requirements for each criterion) using a five-point subjective rating scale (a Likert scale), where 1 indicated the least and 5 indicated the most secured system. We then analyzed the responses and determined a list of the tentative cybersecurity assessment criteria with their relative importance (the total frequency of each criterion proposed by the subjects was to indicate the importance level for that criterion). We also proposed how machine learning and data analytics could be applied to analyze the cybersecurity metrics and enhance cybersecurity in the collaborative system. The proposed cybersecurity metrics can serve as the preliminary effort towards developing comprehensive cybersecurity metrics and methods for human-robot collaborative assembly in automotive manufacturing in particular and for human-robot collaborative systems in general.