Multimodal AI for Intuitive and Safe Control of Upper-Limb Prostheses

Project concept

We are exploring a collaborative research project focused on developing intuitive, adaptive, and safety-aware control interfaces for upper-limb prostheses.

Current prosthetic control systems often rely on myoelectric signals and predefined control strategies. Our proposed research direction investigates how natural-language instructions, electromyography (EMG), and visual perception could be combined to support higher-level, context-aware interaction with prosthetic devices.

The concept builds on ongoing research at the University of Liechtenstein investigating the translation of natural-language commands into structured prosthetic actions with deterministic safety constraints.

Potential research activities

• Multimodal intent recognition integrating natural language, EMG, and visual information

• Vision-language models for object and task understanding

• Translation of user intentions into structured prosthetic control commands

• Safety-aware action planning and execution

• Experimental evaluation of control performance, reliability, and usability

• Integration and validation with robotic or prosthetic hardware

Partners sought

We are particularly interested in collaborating with Polish research groups, universities, and technology partners with expertise in:

• Bionic hands and upper-limb prosthetic systems

• Myoelectric control and biomedical signal processing

• Robotics, intelligent control, and human-robot interaction

• Computer vision and multimodal machine learning

• Rehabilitation engineering and experimental evaluation

Collaboration objective

The aim is to jointly define a technically feasible research and innovation project for the POLNORIS call under the EEA Grants Applied Research and Innovation Programme.

We are seeking a Polish project promoter and complementary research partners interested in developing and evaluating advanced prosthetic control technologies.

The project concept is at an early stage, and the precise research objectives, technical architecture, and experimental activities are open for discussion with potential partners.