Toward Cognitive Augmentation: Personalization and Standardization of Humanoid Social Robot Instruction

Paper in proceeding, 2026

The European industry’s transition towards the human-centric paradigm of Industry 5.0 necessitates the development of advanced training systems that are not only effective but also inclusive. However, current industrial training practices lack a systematic understanding of operators’ diverse needs and often fail to adapt to individual worker characteristics, resulting in standardized approaches that inadequately serve multicultural, multigenerational workforces with varying technical competencies. A primary challenge is creating on-the-job training solutions that adapt to a diverse workforce, comprising individuals with varying cultural backgrounds, technical skills, and cognitive abilities—while minimizing mental workload and accelerating the learning curve. This research addresses this challenge by designing, implementing, and evaluating a humanoid social robot (HSR) demonstrator that provides adaptive, culturally aware, and engaging verbal instructions for industrial tasks. HSR is defined as “human-made technologies that can take physical or digital form, resemble people in form or behavior to some degree, and are designed to interact with people”. It emphasizes that HSRs exhibit both form anthropomorphism (human-like voice or appearance) and behavioral anthropomorphism (gestures, spoken messages, nonverbal expressions). To enhance the social interaction functionality of the humanoid social robot, a natural language model is introduced to enable fluent communication, and computer vision is used to enable eye contact and expression. The HSR’s instructions are conceptualised as the strategy for guidance. We conducted a comparative study between a baseline prototype and an improved version, focusing on four key adaptive dimensions: (1) speech pace control; (2) accent and dialect adaptation; (3) facial expression timing; and (4) instruction granularity. The evaluation will measure usability and instructional efficacy through a combination of quantitative metrics (task completion time, error rate, subjective metrics, and physiological data). The findings of this study imply that personalized HSR interfaces do not lower the mental workload significantly; standardizing HSR interfaces while allowing some flexibility during learning may optimize mental effort and improve the training process. The contribution is a practical, human-centered method to developing adaptive speech interactions with humanoid social robots, with tangible recommendations for inclusive, efficient upskilling in industry.