Research On Smart Wheelchair Interaction Technology Based On Multi-source Information Fusion

The aging of China's population is becoming increasingly serious.At the same time,the number of patients with motor dysfunction caused by accidents is increasing year by year,the demand for mobility equipment has increased.However,most of the existing electric wheelchairs use a rocker control method,which cannot be controlled from a long distance,the interaction method is not intuitive and convenient,and the control method of the wheelchair is also lack of intelligence,which cannot meet the needs of users that grow with the development of technology.Gesture recognition is a novel human-computer interaction method,which has both intuitiveness and convenience.This interaction method is used in a wheelchair interaction control system,which can optimize the human-machine interaction logic and achieve remote control of the device.Gesture interaction can make up for the shortcomings of traditional wheelchair equipment.In addition,in the gesture recognition process,less information is obtained based on a single sensor-based gesture recognition strategy,and the recognition result is susceptible to interference by multiple factors.The recognition system based on multi-source sensor information has better robustness,adaptability and high recognition accuracy.At the same time,the wheelchair interaction scheme that using multi-sensors and intelligent control technology is still in the preliminary research stage,and have better research prospects.Based on this background,this paper proposes research on smart wheelchair interaction technology based on multi-source information fusion.The system is completed by software and hardware circuits.The hardware circuits are divided into the following parts: myoelectric signal conditioning acquisition circuit,acceleration and angle signal acquisition circuit,Bluetooth data transmission circuit,lower computer data analysis receiving circuit and wheelchair DC motor driving circuit.The system can collect multi-source signals from the forearm,transmit the data to the lower computer through Bluetooth,the lower computer receives the data and classifies gestures,and then controls the motor movement mode according to the gesture category.The software part of the system is the gesture recognition algorithm based on multi-source information fusion of the ELM.Firstly,Digitally filter and de-noise the collected data,then a variety of eigenvalues are extracted according to the characteristics of each signal,and the eigenvalues are input into the ELM recognition algorithm.According to the characteristics of various signals,the algorithm adjusts the weights of different actions,builds the recognition algorithm model by using the limit learning mechanism,optimizes the fusion classification structure,and controls the wheelchair movement based on the recognition results.In order to verify the performance of the test system,a variety of tests were carried out,and the following test results were obtained: compared with the single sensor gesture recognition scheme,the recognition accuracy increased by 4% ? 12%,reaching 94.7%;compared with the traditional pattern recognition method,the classification accuracy increased by 1.5% ? 6%,and the online recognition rate of the wheelchair control system reached 95.2%,meeting the real-time requirements of the system.