Myoelectric Control of Powered Upper Limb Prostheses

肢体的丧失或先天性缺失是一种严重的残疾，对截肢者的生活会产生相当大的生理和心理影响。据估计，全球有300多万上肢截肢者，发病率稳步上升。更好的机器人假肢可以大大提高上肢截肢者的生活质量，他们中的许多人拒绝使用现有的设备，因为他们难以像控制完好的手臂那样直观、潜意识地控制它们。 肌电假肢使用来自残肢肌肉的肌电图（EMG）信号（肌肉收缩期间产生的电信号）来控制机动手臂关节。EMG的使用提供了一种建立神经肌肉系统的自然界面以控制肢体丧失的功能的非侵入性手段。 虽然在过去的20年里，在制造更轻、更强和更通用的假体方面取得了重大进展，但在这些假体的可行控制方面几乎没有取得进展。使用过肌电假肢的人清楚地表明，控制的可靠性和灵活性是接受这些设备的最重要因素。 本研究的目的是提供强大的，灵巧的控制肌电假肢。尽管申请人和其他人已经证明了在受控实验室环境中使用基于模式识别和回归的方法在肌电控制中的巨大成功，但是使这些方法在用户的家庭和工作环境中成功的能力需要鲁棒性和灵活性的进步，这将转化为用户体验的有意义的改善。这将通过引入创新的建模和信号处理范例来实现，包括模式识别，自适应学习和非线性回归的新方法。

The loss or congenital absence of a limb is a major disability that can have considerable physical and psychological impact on the life of an amputee. It is estimated that there are more than 3 million upper limb amputees globally, with the rate of incidence growing steadily. Better robotic prostheses can dramatically improve the quality of life for persons with an upper limb amputation, many of whom reject existing devices because they have trouble controlling them in the same intuitive, subconscious way that they controlled their intact arms. Myoelectric prostheses use the electromyogram (EMG) signals (the electrical signals generated during muscle contraction) from residual limb muscles to control motorized arm joints. The use of EMG offers a non-invasive means of establishing a natural interface to the neuromuscular system to control the lost functions of the limb. Although significant advances have been made in building lighter, stronger and more versatile prostheses over the last 20 years, little progress has been made in viable control of these prostheses. Individuals who have used myoelectric prostheses clearly indicate that it is the reliability and dexterity of control that is the most significant factor in acceptance of these devices. The objective of this research is to deliver robust, dexterous control to myoelectric prostheses. Although the applicant, and others, have demonstrated great success in myoelectric control using pattern recognition and regression based methods in controlled laboratory settings, the ability to have these methods succeed in user's homes and work environments requires advances in robustness and dexterity that will translate into a meaningful improvement in user experience. This will be accomplished by introducing innovative modeling and signal processing paradigms, including novel methods in pattern recognition, adaptive learning, and nonlinear regression.