Neural networks for control of posture and locomotion

摘要

Neural network algorithms have recently been used in a wide variety of control systems applications. In some applications, neural network components are used in place of other control system components in a standard control system configuration-e.g., neural networks have been used as inverse system models in direct inverse control. In other applications, neural network components have been used in somewhat nonstandard control system configurations, such as configurations that are based on the structure of neurophysiological control systems. An important example is the use of neural network pattern generators for the control of locomotion. This paper presents a review the author's work and the work of others in applying neural networks for the control of posture and locomotion. In the author's work, he is using neural networks to control locomotion and posture in individuals with spinal cord injury using functional neuromuscular stimulation. The author's approach is similar to that of others whose applications cover a range of activities from basic science investigations of motor control systems to controlling locomotion of hexapod robots.