 From Fragmentation to Integration: Designing Humanoid Robots for Corporate Service EnvironmentsMarcel Leichtle Publications of Darmstadt Technical University, Institute for Business Studies (BWL) from Darmstadt Technical University, Department of Business Administration, Economics and Law, Institute for Business Studies (BWL) Abstract: Humanoid and anthropomorphic robots – embodied forms of artificial intelligence (AI) that can move, perceive, and converse – are increasingly entering organizational settings, but their value hinges less on raw capability than on disciplined alignment of form, interaction, and governance under real enterprise constraints. The managerial question is not whether adoption will occur, but how to specify and integrate these systems so that the promise implied by human-like morphology can be delivered reliably in routine work. This dissertation responds by proposing an integrated, three-level socio-technical framework that, while not claiming to eliminate technical limits or expand robots into wholly new domains, instead, structures trade-offs, reduces fragmentation between engineering and service design, and embeds conversational intelligence within auditable controls. The framework organizes design and deployment around (i) physical embodiment, (ii) socio-technical design principles, and (iii) intelligent ecosystem integration. It advances a structured understanding of robot design, derives validated organizational requirements (via echeloned Design Science Research; eDSR), and operationalizes conversational intelligence with a novel Knowledge Interaction Distillation (KID) pipeline within enterprise ecosystems. This dissertation presents three studies: Study I develops a design-oriented map of the anthropomorphic robot landscape. Using morphological analysis, it organizes the design space across five interdependent dimensions – appearance and human-likeness, mobility and locomotion, human–robot interaction (HRI) modalities, construction and modularity, and sensors for audio, vision, and touch. Applying this morphological box to current and emerging platforms reveals systematic clustering: many robots are non-modular, carry limited loads, and rely on narrow HRI repertoires, which helps explain their difficulty transitioning from pilot to production. The analysis argues for a so defined gold-standard orientation that elevates modular construction, richer sensing, and broader interaction skills alongside competent manipulation and mobility. Beyond cataloguing features, the study positions morphological analysis within Design Science Research to prevent fragmented decisions and to make trade-offs explicit for both engineers and managers. Conceptually, this establishes Level 1 (Physical Embodiment) and provides a standardized vocabulary and scorecard for procurement and portfolio steering. Building on this foundation, study II addresses the translation problem from aspiration to implementable requirements. Adopting eDSR, it combines a systematic literature review (SLR; n = 676) with a year-long feasibility collaboration to derive a concise set of design objectives for enterprise-ready humanoid systems. The resulting objectives – simplicity for maintainability and training, modularity for flexibility and lifecycle cost, autonomy for reliable behavior in dynamic settings, interaction for intuitive, multi-modal use, and adaptability for sustained value creation – are each expressed in paired conceptual and technical requirements and validated with domain experts. This stage-gated approach keeps service experience and robotics engineering aligned, turning generic goals into organization-ready commitments that can guide architecture, procurement, and governance. This constitutes Level 2 (Socio-Technical Design Principles), converting the descriptive map into stage-gated, testable requirements that align service experience with robotics engineering. Study III turns to the intelligence layer and examines how conversational capability can be made robust, auditable, and efficient under organizational constraints. It introduces KID, a participatory, simulation-driven method that transforms interaction design into compact, controllable language models. Through multi-agent simulations that encode user personas, system logic, and evaluators, the method generates and filters synthetic dialogues to fine-tune smaller models suitable for local or tightly governed hosting. In a year-long action-research collaboration with a small-to-medium enterprise, this approach powers a distributed assistant that runs across a humanoid platform and a companion mobile application. Empirically, the distilled models outperform prompt-only baselines on constraint following and output consistency while meeting privacy and latency demands. Practically, the work reframes the ‘robot’s intelligence’ as ecosystem capability: the same interaction logic drives the co-present robot and the user’s personal device, and modality choices, such as avoiding voice input in open offices, are grounded in observed behavior rather than assumptions. This realizes Level 3 (Intelligent Ecosystem Integration) as a shared, policy-aligned intelligence layer that behaves consistently across embodiments and survives enterprise governance. In summary, this dissertation advances theory by weaving morphology, design science, and conversational-AI customization into a unified account of deployable humanoid systems. Methodologically, it offers reusable instruments – a morphological box for anthropomorphic robots, a stage-gated requirements process, and a participatory distillation pipeline – that future researchers can adapt. For managers, it provides practical guidance on platform selection, modular architecture, and governance-friendly customization of conversational intelligence. In sum, it argues that embodied AI will ‘work at work’ only when physical design (Level 1), validated socio-technical requirements (Level 2), and operational intelligence and ecosystem integration (Level 3) are engineered together, turning robots with a face and a voice into reliable teammates who fit the workplace. Date: 2026-05-21 Downloads: (external link) Related works: Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text Persistent link: https://EconPapers.repec.org/RePEc:dar:wpaper:160982 Access Statistics for this paper More papers in Publications of Darmstadt Technical University, Institute for Business Studies (BWL) from Darmstadt Technical University, Department of Business Administration, Economics and Law, Institute for Business Studies (BWL) Contact information at EDIRC. |
Is your work missing from RePEc? Here is how to contribute.
Questions or problems? Check the EconPapers FAQ or send mail to .
EconPapers is hosted by the
School of Business at Örebro University.