This article reviews technologies and algorithms for decoding volitional movement intent using bioelectrical signals recorded from the human body. Such signals include electromyograms, electroencephalograms, electrocorticograms, intracortical recordings, and electroneurograms. After reviewing signal features commonly used for interpreting movement intent, this article describes traditional movement decoders based on Kalman filters (KFs) and machine learning (ML). A number of deficiencies of the current state of the art in this field are described, and three approaches that mitigate some of these deficiencies are reviewed. They include data aggregation-based training to improve decoder performance when only limited amounts of training data are available, a shared controller that incorporates estimates of movement goals, and an adaptive decoder designed to compensate for time variations in the relationships between the human body and the prosthesis. Also included are experimental results that illustrate some of the concepts discussed in the article.

中文翻译：

---

从生物信号解释意志运动意图：综述

本文回顾了使用从人体记录的生物电信号解码意志运动意图的技术和算法。此类信号包括肌电图、脑电图、皮层电图、皮层内记录和神经电图。在回顾了通常用于解释运动意图的信号特征之后，本文描述了基于卡尔曼滤波器 (KF) 和机器学习 (ML) 的传统运动解码器。描述了该领域当前技术水平的许多缺陷，并审查了减轻其中一些缺陷的三种方法。它们包括基于数据聚合的训练，以在只有有限数量的训练数据可用时提高解码器性能，一个包含运动目标估计的共享控制器，以及一个自适应解码器，旨在补偿人体和假肢之间关系的时间变化。还包括说明文章中讨论的一些概念的实验结果。