23. Juli 2026 | Tübingen

Replicating the complexity of human movement

Cyber Valley Research Fund project teaches AI systems to move more like humans

Daniel Häufle, Georg Martius, and Pierre Schumacher have completed their project “Learning efficient control of non-linear muscle-driven systems: Morphological computation as guiding principle”. The project was funded by the Cyber Valley Research Fund and was carried out between 2021 and 2024 at the University of Tübingen. Its aim was to test whether morphological computation can improve how computer models of the human body learn to replicate its movements.

Morphological computation (MC) is the process by which the body moves and interacts with the world without relying on the brain to calculate the exact forces that need to be applied by each muscle and joint. For example, the brain only needs to give the simple instruction “close your hand”, while low-level reflex arcs, in-built neural architectures, and muscle properties automatically coordinate the exact forces, timing, and adjustments needed to successfully perform the movement.

The researchers first developed a new learning approach that allows computer models of the human body with many muscles to learn complex movements more effectively. They found that having many muscles makes learning harder, but by guiding the learning process more carefully, they were able to produce stable, human-like movements such as walking, even across uneven terrain. Building on this, they showed that adding simple, muscle-like properties makes movements more robust and natural, both in simulation and on real robots. Remarkably, a walking pattern learned in simulation worked on a real robot without extra training.

The project demonstrated that morphological computation can advance computer models of the human body. The research represents the first step towards bringing the full diversity of human movement to artificial systems and robotics. With potentially broad applications across various industries, such as healthcare, rehabilitation, prosthetics, and advanced manufacturing, it aligns with Cyber Valley’s mission to translate cutting-edge research into real-world applications for a better future.

This project produced the following peer-reviewed publications:

DEP-RL: Embodied Exploration for Reinforcement Learning in Overactuated and Musculoskeletal Systems (ICLR 2023): https://openreview.net/forum?id=C-xa_D3oTj6
Emergence of natural and robust bipedal walking by learning from biologically plausible objectives (iScience 2025): https://doi.org/10.1016/j.isci.2025.112203
Learning with Muscles: Benefits for Data-Efficiency and Robustness in Anthropomorphic Tasks (CoRL 2022): https://openreview.net/forum?id=Xo3eOibXCQ8
Learning to Control Emulated Muscles in Real Robots: A Software Test Bed for Bio-Inspired Actuators in Hardware (BioRob 2024): https://doi.org/10.1109/BioRob60516.2024.10719699

About the Cyber Valley Research Fund

The Cyber Valley Research Fund was established to support Cyber Valley research groups undertake basic research in the fields of artificial intelligence and robotics. The fund totaled five million euros, including contributions from six of Cyber Valley’s founding corporate partners: Amazon, BMW, Bosch, IAV, Mercedes-Benz, Porsche, and ZF. It supported 20 research projects, the first of which began in 2020, and the final of which will conclude in 2026.