Stair climbing and descending phase recognition method and system based on random forest and computer readable medium

CN117257617BActive Publication Date: 2026-07-07MEBOTX INTELLIGENT TECH SUZHOU CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
MEBOTX INTELLIGENT TECH SUZHOU CO LTD
Filing Date
2023-09-20
Publication Date
2026-07-07

AI Technical Summary

Technical Problem

In existing technologies, exoskeleton robots suffer from poor sensor durability when recognizing gait patterns when going up and down stairs, especially the foot sensors, which are prone to damage. Furthermore, the gait recognition results are inaccurate and they are difficult to adapt to abnormal gait patterns. Traditional machine learning methods rely on foot pressure sensors, which leads to instability in daily life.

Method used

A random forest-based stair assist phase recognition method is adopted. Data is collected by IMU sensors in the thigh and calf positions to construct a training set and perform dimensionless preprocessing. The random forest classifier is used to train the assist phase recognition model, removing the dependence on the foot sensor and relying solely on the IMU sensors in the thigh and calf for recognition.

Benefits of technology

It improves the accuracy and stability of gait recognition, reduces sensor costs, avoids sensor damage, saves data annotation time and manpower and resources, and achieves a recognition accuracy rate of over 98%.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117257617B_ABST
    Figure CN117257617B_ABST
Patent Text Reader

Abstract

The application provides a stair climbing and descending assistance phase recognition method, system and computer readable medium based on a random forest, which comprises the following steps: acquiring wearing sensor data of different human bodies in the process of performing stair climbing and descending actions and non-stair climbing and descending actions; determining the foot movement state according to the plantar IMU sensor data, determining the movement gait phase and labeling; constructing a training set by using the IMU sensor data of the thigh and calf positions in the wearing sensor data and the labeled movement state; training a stair climbing and descending assistance phase recognition model based on a random forest classifier; and finally, determining the IMU sensor data of the thigh and calf positions in the actual acquired human body action by using the assistance phase recognition model, and outputting the corresponding movement assistance state. The method of the application can be used for recognizing the assistance phase in the process of stair climbing and descending, has a simple sensor layout, saves the time consumed in the data labeling process, and can improve the recognition accuracy.
Need to check novelty before this filing date? Find Prior Art