PF-HCPS

Background: Human–cyber–physical systems (HCPS), as a natural extension of cyber–physical systems (CPS), add considerations of human interaction and cooperation with systems and the physical world, human and system as a team to achieve target goals through HCPS collaboration. Humans are a part of the system in HCPS and are contained within the system boundary. The complex human–cyber–physical collaboration patterns exhibit distinct states in varying environments, interacting with their surroundings differently. The uncertainty of human behavior and environments in HCPS makes modeling and analyzing HCPS more challenging. Motivation: Problem frames (PF) provide a way to describe the domain problems or requirements to help developers understand the context in which the software problem resides. In contrast to other approaches, PF emphasizes the characterization of the real world in which the software system will act and refers to the meaning of requirements to the description of the relevant domain in the real world. However, PF cannot zoom in on the details in the problem domain to describe the human–machine collaboration patterns, which are the core concepts of HCPS. Method: In this paper, we extend problem frames (PFs) named PF-HCPS to support modeling and directly analyzing human–cyber–physical entities and their collaborations. We introduce system boundaries into the PF model and split domain elements into system and environment categories. Then, we introduce a system-level state machine to specify the patterns of human–machine collaborations and a concept-level state machine to describe domain elements. Evaluation: We use the Tesla Autopilot Model S as the case study to demonstrate our proposed model. As one of the most widely used HCPSs, the Tesla Autopilot system has a high level of complexity and relatively complete instructions for use, which contain 17 Autopilot-related function points and 8 collaboration patterns. Overall, the results were satisfactory. PF-HCPS can cover all the functional requirements and the human–machine collaboration patterns. The proposed approach can be further extended and applied to environment-driven requirements engineering in the software industry. Conclusion: PF-HCPS is proposed to support HCPS collaboration. Overall, the results were satisfactory. PF-HCPS enables the complete description of HCPS requirements and can characterize human–machine collaboration patterns in domains. The proposed approach can be further extended and applied to environment-driven requirements engineering in the software industry.