Development of a Programmable Upper-Body Exoskeleton with AI-based Motion Capture and Feedback (ExoTrain)

Motion Capture and Support for learning and training of movements

Exoskeleton and Feedback supported training

Project description

Ramblr GmbH and Technische Universität Chemnitz (Professur Produktionssysteme und -prozesse) are working together to make complex movements easier to learn and train by combining AI-based motion analysis, digital twins, and wearable robotics.

ExoTrain develops a programmable upper-body exoskeleton, from shoulder to wrist, that enables users to learn and practice movements from video examples. AI extracts reference movements from 2D videos and adapts them to the trainee’s body. A digital twin compares these movements with the actual execution in real time and shows any deviations, while the exoskeleton corrects them through vibration and motorized guidance.

While Ramblr develops the AI pipeline for skeleton extraction, skeleton morphing, and motion transfer, TU Chemnitz designs, builds, and validates the exoskeleton and its control software. The university also develops the interaction, feedback, and training system with the digital twin. A user-centered approach involves trainers and future user groups in defining the requirements and use cases, while user studies help shape the interaction and evaluate the overall system.

ExoTrain creates new possibilities for guided movement training, from unsupervised exercise at home and data-driven gym workouts to learning sports techniques, gaming, and immersive interaction in virtual reality.

Project data

Project period: Oct. 2026 – Mar. 2029

Coordinator: ramblr.ai GmbH

Project partners

Funding partner