TEAM-AI: Trustworthy Embodied Assistance in Manufacturing with AI

This KIEM Arbeidsbesparende AI project investigates the feasibility of deploying an AI-enabled humanoid “buddy” robot to support workers in SME manufacturing environments. In factories, skilled operators often spend considerable time on repetitive support-tasks such as fetching tools, transporting components, or navigating frequently changing layouts. These inefficiencies increase physical strain and cognitive workload while reducing productivity.
TEAM-AI is initiated in response to a direct demand from our industrial partners to investigate the potential of humanoid robots, leveraging existing AI-based perception and navigation solutions, to assist workers in SME manufacturing environments. The main research question focuses on the application and validation of existing AI methods for environment perception, sensor fusion, and human-aware navigation to enable safe human–humanoid collaboration. This includes evaluating proven AI models for real-time object detection, mapping, localization, and motion planning, as well as defining measurable criteria for productivity performance, safety, and scalability.
This study targets practical use cases such as on-demand tool transport, adaptive workspace mapping, and support for dynamic order picking in industrial manufacturing environments of our partners. By testing these solutions within partner-driven use cases and realistic workflows, the project generates insights into their operational value, implementation barriers, and conditions for successful adoption. Moreover, it advances trustworthy humanoid assistance systems that enhance worker well-being while preserving industrial flexibility.
The one-year feasibility study combines requirement analysis with industrial partners, the integration and configuration of existing AI-based functionalities, iterative task testing with workers, and structured evaluation of safety and operational impact. No new AI methods will be developed; instead, the project focuses on the practical implementation and validation of proven AI techniques for enabling a humanoid robot in an industrial context. Outcomes include validated demonstrators in real-life SME settings, quantified feasibility metrics, and a roadmap toward scalable deployment and follow-on industrial innovation projects.