Researchers from the University of Granada (UGR) and the École Polytechnique Fédérale de Lausanne (Switzerland) are developing breakthrough that could have implications in healthcare, manufacturing and agriculture, among other sectors.
Scientists from the UGR’s Department of Computer Engineering, Automation and Robotics (ICAR) are leading the design of a controller that combines a computational model of the cerebellum with a computational muscular model to coordinate the movements of a robotic arm.
The study combines neuroscience and biomechanics to develop a controller capable of adjusting a robot’s motor behaviour. The objective is to enable robots to adapt their actions to the requirements of the environment in which they are operating and the tasks they are performing.
“Lifting a flower pot is not the same as handling an egg. The first task requires rigidity in our joints, while the second requires much smoother movements. Thanks to the combination of the nervous system and the biomechanics of the body, we can adjust our movements according to the context and cover a broad range of motor behaviours”, explains principal investigator Ignacio Abadía Tercedor. Robotics research aims precisely to replicate this versatile behaviour.
The cerebullum model developed by the researchers makes it possible to learn how to control the robot’s arm. “Just as we learn to coordinate our bodies from an early age, the cerebellum model starts from scratch and gradually learns to control the robotic arm so that it can perform different movements,” Ignacio Abadía adds.
Meanwhile, the muscle prototype replicates the viscoelastic properties that are characteristic of muscle biomechanics, and also includes variable co-contraction and a spinal cord reflex response. Adjusting muscle co-contraction (the simultaneous activation of the agonist and antagonist muscles that move a joint) changes the stiffness of the robot’s arm, just like movement with our joints.
By combining both models, the researchers are able to make the robotic arm adjust its behaviour according to the context: if greater precision and rigidity are required, the robot applies greater co-contraction; if smoother movements are required — for example to interact safely with humans — the robot minimises rigidity using low levels of co-contraction.
Ignacio Abadía, Eduardo Ros and Niceto R. Luque, all of whom are participating in the study, are researchers from the UGR’s School of Computer and Telecommunication Engineering (ETSIIT).
Bibliographic reference:
Ignacio Abadía et al. A neuromechanics solution for adjustable robot compliance and accuracy. Science Robotics. 10, eadp 2356 (2025). 10.1126/scirobotics.adp2356
Contact details:
Ignacio Abadía Tercedor
Department of Computer Engineering, Automation and Robotics (ICAR)
School of Computer and Telecommunication Engineering (ETSIIT)
University of Granada
Email: iabadia@ugr.es
Translated version: This text has been translated into English by the Language Services Unit (Vice-Rectorate for Internationalization) of the University of Granada.
