Gait Assistance System

Gait Assistance System Predicting Ankle Stiffness via EMG-based Machine Learning

This technology is a gait assistance system that calculates the predicted stiffness of the ankle joint by analyzing the wearer's lower limb electromyography (EMG) signals using LSTM or CNN-based machine learning algorithms, and controls the ankle joint angle by integrating data from foot contact sensors.

Existing stiffness prediction methods using muscle models or mathematical linearization formulas have suffered from low accuracy, as they fail to fully account for the non-linear characteristics and time-varying states of muscles.

This technology proposes a method that trains machine learning algorithms based on EMG and stiffness data from multiple pedestrians to predict real-time stiffness, while utilizing combined data from toe and heel contact sensors to calculate target angles. It can be applied to rehabilitation therapy and gait assistance for the elderly, providing natural gait support by adapting in real-time to changes in the wearer's muscle condition.

‍

‍

Key Features:

• A gait assistance robot worn on the user's foot to assist walking by adjusting ankle joint stiffness
• A plurality of EMG sensors attached to muscles associated with ankle movement to detect electromyography signals
• A stiffness prediction unit that receives EMG signals and outputs predicted stiffness through a machine learning algorithm
• A robot control unit that regulates the ankle joint stiffness of the gait assistance robot based on the predicted stiffness

‍

This invention was developed with support from the Ministry of Culture, Sports and Tourism's project for hybrid smart clothing and monitoring systems for enhanced athletic performance, and the Ministry of Science, ICT and Future Planning's Human-Centered Soft Robot Technology Research Center.