基于视觉和表面肌电信号的关节角度估计方法及装置
By combining a multimodal approach using monocular vision and surface electromyography signals, and utilizing an LSTM model for joint angle estimation, the problems of low signal-to-noise ratio and insensitivity to minute arm movements in existing technologies are solved, achieving higher accuracy and more stable joint angle estimation.
CN119184673BActive Publication Date: 2026-07-17DALIAN MARITIME UNIVERSITY
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- DALIAN MARITIME UNIVERSITY
- Filing Date
- 2024-11-05
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
In the existing technology, the joint angle estimation method based on surface electromyography signals has low signal-to-noise ratio and is not sensitive to small arm movements, resulting in insufficient estimation accuracy and stability.
Method used
By combining monocular vision and surface electromyography (EMG) signals, a camera captures motion video streams and extracts visual features, while an EMG sensor captures EMG signal streams. An LSTM model is then used for multimodal feature fusion and prediction to achieve continuous estimation of joint angles.
Benefits of technology
It improves the accuracy and stability of joint angle estimation and enhances the model's ability to continuously estimate joint angles.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN119184673B_ABST
Abstract
本发明公开了一种基于视觉和表面肌电信号的关节角度估计方法及装置，方法包括：捕捉动作视频流；捕捉肌电信号流；根据动作视频流提取初级三通道视觉特征；根据肌电信号流捕捉肌电特征；对初级三通道视觉特征进行超采样，从而获取与肌电特征等长度的二级三通道视觉特征；将所述二级三通道视觉特征和肌电特征输入训练后的LSTM模型中进行预测，所述训练后的LSTM模型用于将内部参数映射到肩、肘、腕三个关节角度；获取所述训练后的LSTM模型的输出作为肩、肘、腕三个关节角度的预测值。本发明研究了多模态感知数据对连续手臂关节角度估计的影响，能够提高基于表面肌肉电信号对关节角度连续估计的精度和稳定性。
Need to check novelty before this filing date? Find Prior Art