Pneumatic Muscle Control via EMG Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for controlling pneumatic muscle devices in exoskeletons rely on preset action signals, leading to significant hysteresis in muscle contraction and reduced accuracy and user experience.
Innovation Solution
A method and device that utilize electromyographic signals to control pneumatic muscle devices, involving the acquisition of electromyographic signals from target muscles, classification of user behavior using a trained neural network model, determination of driving amounts for each muscle, and driving of simulated muscles based on these amounts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If preset action signals are used to control pneumatic muscles, then the control system is simple, but significant hysteresis occurs in muscle contraction compared to muscle activation intention
Solution Approach 1:
The patent employs feedback by acquiring electromyographic (EMG) signals from the user's muscles and using these signals to control the pneumatic muscles. The EMG signals provide real-time information about the user's muscle activation state, allowing the system to adjust the pneumatic muscle contraction accordingly, thereby eliminating the hysteresis present in preset control systems while maintaining relatively simple device architecture.
2Ease of operation
If preset action signals are used to control pneumatic muscles, then the control method is easy to implement, but the accuracy and personalization of the control are reduced
Solution Approach 1:
The system utilizes the user's own EMG signals as the control input, allowing the control method to be personalized to each user's natural muscle activation patterns. This self-service approach enables accurate and personalized control without requiring complex programming or preset configurations, as the system adapts to the user's inherent control characteristics.
3Measurement precision
If EMG signals and neural network model are used to control pneumatic muscles, then the control accuracy and personalization are improved, but the device complexity increases
Solution Approach 1:
The patent introduces an intermediary processing layer that includes EMG signal acquisition, preprocessing, and neural network-based classification. This intermediary system translates complex physiological signals into control commands for the pneumatic muscles, achieving high accuracy and personalization while keeping the overall system architecture manageable through modular design.
Data Source
AI summary
A method of controlling a pneumatic muscle device through an electromyographic signal including: obtaining electromyographic signals currently generated by each of the target muscles of the wearer of the pneumatic muscle device; determining a behavior classification of the wearer of the pneumatic muscle device according to all of the electromyographic signals and a preset neural network model, and the neural network model is trained and obtained according to historical electromyographic signals generated by each of the target muscles; determining a driving amount corresponding to each of the target muscles according to the behavior classification of the wearer of the pneumatic muscle device; and driving the simulated muscles corresponding to each of the target muscles according to the driving amount corresponding to each of target muscles.