Description | Please join the Center for Sensorimotor Neural Engineering (CSNE) in welcoming Dr. Francisco Valero-Cuevas to Seattle. Dr. Valero-Cuevas will present a CSNE/Kavli seminar titled “Reverse-engineering Spinal Circuitry". Dr. Valero-Cuevas is the Director of the Brain-Body Dynamics Laboratory, Professor of Biomedical Engineering, Professor of Biokinesiology and Physical Therapy, Professor of Computer Science & Aerospace and Mechanical Engineering, University of Southern California. Seminar Abstract: The objective of this work is to build a neuromorphic robotic system that can interact with the physical world by implementing neuromechanical principles. It is a faithful implementation of the spinal circuitry responsible for the afferentation of muscles and is capable of producing both normal and pathological functions. We used state-of-the-art models of muscle spindle mechanoreceptors with fusimotor drive, monosynaptic circuitry of the stretch reflex, and alpha motoneuron recruitment and rate coding. This multi-scale, hybrid system driven by populations of 1024 spiking neurons, emulated the physiological characteristics of the afferented mammalian muscles. We implemented these models on field-programmable gate arrays (FPGAs) which are capable of running these complex computations in real-time. The FPGAs control the forces of two muscles acting on a joint via long tendons. We performed ramp-and-hold perturbations and systematically explored a range of muscle spindle gains (fusimotor drive) to characterize the stretch reflex response in different phases of the perturbation. Finally, we explored the fidelity of four models for isometric muscle force production by testing their responses to rate-coding using spike trains and produced force ramps. This autonomous integrated system was self-stable and the closed- loop behavior of populations of muscle spindles, alpha and gamma motoneurones, and muscle fibers emulated muscle tone and function. Sweeping the range of muscle spindle gains provided us with a subset of values that produced tenable physiological and pathological responses. Moreover, isometric force generation revealed that the dynamic response in the tendons is very sensitive to tendon elasticity, especially at high firing rates. This hybrid, neuromorphic, neuromechanical system is a precursor to neuromorphic robotic systems. It provides a platform to study healthy function and the potential spinal and/or supraspinal sources of pathologic behavior. |
---|