Reciprocal learning in human–machine collaboration: a systematic literature review and implications for production and logistics

Abstract

The integration of automation technologies and artificial intelligence into production and logistics is transforming work organisation for human and machine agents. Beyond the scope of classical collaborative task allocation problems, studies on social aspects, especially the mutual learning of humans and intelligent machines during interactions, remain scarce. By enhancing cyber-physical production and logistics systems to become intelligent, learnable, and social, human–machine symbiosis can be fostered to enhance their complementary strengths. Despite studies addressing the potential of this bidirectional learning process in the form of reciprocal human–machine learning (RHML), this concept remains ambiguous and lacks a comprehensive knowledge base in production and logistics. Therefore, in this study, a systematic literature review was conducted to gather and categorise the existing knowledge on RHML in different disciplines. Further, efforts were made to (i) consolidate existing design components of RHML into a taxonomy; (ii) describe current design patterns of RHML with classified RHML archetypes; and (iii) apply the resulting taxonomy and archetypes to discuss the potential of RHML concepts in production and logistics. This interdisciplinary approach aims to extend the existing design concepts in cyber-physical production and logistics systems. In addition, initial discussions on the future research agenda are provided.