Talk: From Human Modeling to Neurally-Driven Symbiotic Embodiment

When: 25.04.24
Where: GLC E 34.1
What: 

Title: From Human Modeling to Neurally-Driven Symbiotic Embodiment
Speaker: Prof. Sami Haddadin, TU München, Chair for Robotics and System Intelligence

Abstract: In the development of transformative neuroengineering technologies for the restoration of autonomy and mobility to those living with physical disabilities, a shift towards a more compatible, safe, and anticipatory human-centered approach is essential. This necessitates the integration of advanced AI and machine learning algorithms capable of leveraging high throughput sensory data (including kinematics, eye-tracking, invasive and non-invasive brain data) and human neuromechanics models (human digital twins). Rather than expecting humans to conform to and learn how to operate robotic systems, including prosthetics, robot neuroassistants or brain-controlled humanoids, the focus is on designing systems that recognize human intent and emulate natural limb and whole body behavior. This entails embedding models of human movement and interaction into robot learning and control algorithms, maximizing therapeutic effects and patient benefit as well as achievable autonomy.

The core of this approach is the development of human digital twins driven by multi-modal sensory data, encompassing neuromechanics modeling, muscle dynamics, and control mechanisms. By learning from human physiology, robot systems can exhibit anthropomorphic traits and reflexes, achieving advanced capabilities while retaining human-like responses. The use of intelligent control and learning algorithms enables the realization of AI-driven systems, such as prostheses, which facilitate real-time, user-in-loop interaction and functionality such as grasping and motion compensation.

Under this new perspective, transformative technologies also advance beyond basic Brain-Machine Interfaces (BMI), with the focus shifting from mere brain signal processing to neurally-driven symbiotic embodiments, aiming to decode neural signals for more natural human-like behavior in robotic systems. This includes decoding full state joint, impedance, and force policies and integrating human digital twins to enhance user embodiment.

In summary, this new generation of transformative technologies paves the way for a new era of human-model-informed, symbiotic embodiments, enhancing the integration and effectiveness of AI-empowered robotic systems in various domains, particularly in healthcare and rehabilitation.

Bio:                                                                                                                                                                                    Prof. Haddadin is currently one of the most renowned researchers in robotics, AI applied to rehabilitation and geriatrics. He is the inventor and founder of Franka Emika GmbH, which sold thousands of its robots to industry and academia. He has received the Leibniz-Preis from the DFG, the most prestigious German research award, and the “Deutscher Zukunftspreis» from the German Bundespräsident. He currently building a healthcare and robotics center in Garmisch-Partenkirchen, DE, one of the largest facilities of its kind in the world.