Real-time prediction learning for the simultaneous actuation of multiple prosthetic joints

摘要: Integrating learned predictions into a prosthetic control system promises to enhance multi-joint prosthesis use by amputees. In this article, we present preliminary study of different cases where it may be beneficial set temporally extended - and maintained in real time within an engineered or controller. Our demonstrates the first successful combination actor-critic reinforcement learning with real-time prediction learning. We evaluate new approach during myoelectric operation robot limb. results suggest that integration speed policy acquisition, allow unsupervised adaptation controllers, facilitate synergies highly actuated limbs. These experiments also show enables anticipatory actuation, opening way for coordinated motion assistive robotic devices. work therefore provides initial evidence realtime is practical support intuitive joint increasingly complex systems.