Abstract (short version):
In sectors with low standardization of processes or products and characterized by dynamic and complicated environments, processes are still mostly manual. For non-delocalizable activities, such as logistics or healthcare, this leads to productivity bottlenecks, especially in those countries with aging population. The automation of such non-delocalizable yet challenging processes is therefore crucial for the competitiveness of industrialized countries. However, since the dynamicity and complexity of such processes make a full automation hardly achievable, the integration of human operators with automated agents in socio-technical system becomes a necessary compromise. While designing such human-automation interaction systems, the most challenging question to answer is, what should be automated and what should be carried out manually. In this thesis work, a method is presented – the Quality Interaction Function Deployment – that enables the selective automation of functions necessary to accomplish a process. By taking into account at the design phase requirements of several groups of stakeholders of the automation, the method warrants focusing on what is important to automate, yielding the best value for automation.
Abstract (long version):
In sectors with high process and product/service standardization, the cost-benefit ratio of physical and digital automation solutions is high as they contribute to increase capacity, decrease costs and improve product/service quality. In less standardized fields of application, the contribution of physical or digital process automation to improvements in capacity, cost and quality blurs, because the automation of complex and unstructured processes requires sophisticated, flexible, thus expensive systems. Due to the complexity and lack of standardization, such automated systems might still be low-performing (physical processes) or fragile (digital processes), which questions the overall impact on the product/service quality as perceived by the end customer. Yet, the sole consideration of context-dependent economic parameters leads to pursuing full automation of physical or digital process chains even in automation-averse sectors. This often results in overall sub-optimal solutions, in which the final socio-technical systems do not behave as designed due to the lack of consideration of human factors. Taking the distance from the false idea that a process should be either fully automated or fully manual, this thesis work aims to the development of a method enabling the selective automation of functions necessary to accomplish a process, be that of digital or physical nature. The method will consider at design phase stakeholders’ requirements of different nature. First, the system requirements representing the view of the investors will be consider, enabling the consideration of context-dependent operational aspects. Then, using several quantitative indicators, a heuristic approach for prioritizing the allocation of the identified functions to an automated agent will enable the creation of few automation scenarios. Such function allocation corresponds to the design of an engineering hypothesis for each automation scenarios. Finally, requirements depending on human nature rather than a specific economic context will be considered in the design phase for the evaluation of designed engineering hypotheses. Among the human factors requirements considered at the design phase, we will consider ergonomics, system complexity, workload balance, accountability and acceptance. This novelty enables taking into consideration aspects implicitly or explicitly important for users and operators already at design phase, before realization. In turn, this means that, thanks to a thoroughly evaluated allocation of functions among human and automated agents, the resulting designed socio-technical systems will be more likely to achieve the target performances when implemented in reality. Thanks to the variables and mechanisms embedded in the method, high compliance with both sets of requirements (i.e. context-dependent and context-independent) will become a proxy for good utilization of the synergies of human-robot interaction. As such, the method will warrant focusing on what is important to automate, yielding the best value for automation. In analogy with the principles of waste elimination and value enhancement of lean thinking and lean manufacturing, the method will enable the design of lean human-robot interaction, as mentioned in the title of the work.