Analysis And Prediction Of Indoor Pedestrian Trajectory Based On Machine Learning Algorithms

With the advent of the field of communication internet and artificial intelligence,the total amount of data in the world is increasing exponentially.Therefore,the signal accumulated data generated in the communication internet is also faced with problems such as how to store,analyze,process and apply.Based on the above massive data,the traditional data processing methods are no longer applicable.At present,big data processing and analysis and artificial intelligence(machine learning algorithm and neural network technology)are becoming more and more mature and applied in all walks of life.Nowadays,indoor pedestrian positioning,action trajectory estimation and prediction is an important direction based on the integration of wireless signal transmission,big data processing and artificial intelligence.The reason is that there is a lack of satellite signal or weak satellite signal in shopping malls with high buildings and fire sites,Pedestrian positioning and trajectory prediction can effectively improve the judgment of the accurate position of people and the future route.In this way,it is helpful to judge the position of pedestrians in real time and provide better services in shopping malls,or assist rescue personnel to judge the position of trapped personnel at the disaster site,so as to track their movements.In recent years,indoor pedestrian trace analysis and prediction continue to rise.This field involves a series of complex links such as signal data acquisition,storage,processing,analysis,modeling,testing and result prediction.Researchers at home and abroad have put forward various solutions and related technologies for each link,but there is a lack of complete and effective mechanism for the whole process processing.Therefore,based on the inertial sensor fixed on the walking personnel,this paper measures the pedestrian motion data in real time,transmits the data to the background system for storage and sorting,processes the corresponding data(such as smoothing,denoising,blank filling value,etc.)using big data technology,and calls the corresponding machine learning model and algorithm to model the processed data,Train and learn the existing data,save the trained model,fit the new pedestrian travel data,and then predict the future trajectory of pedestrians.This paper mainly completes the following work:(1)Acquisition,transmission and storage of indoor personnel movement data based on Bluetooth:Considering that the shoulder position is relatively stable in the process of walking,this paper binds the inertial sensor to the shoulder position of the person,and transmits the acceleration and steering data of the pedestrian in the process of walking to the mobile phone in real time and stores it based on Bluetooth technology.(2)Processing and analysis of static,straight-line and rotating data of indoor personnel:Due to the influence of various factors such as noise in the transmission process of signal data,it is necessary to process the signal data accordingly to achieve the data availability level.This paper will use big data processing and analysis technology to denoise,smooth,fill in vacancy values,delete and so on.(3)The pedestrian behavior trajectory is modeled and the travel position of future personnel is predicted:Model the pedestrian behavior trajectory and predict the future travel position of personnel.For the processed data,use machine learning algorithms(mainly logical regression and random forest algorithm)to model the pedestrian behavior trajectory,predict the future travel position of personnel,test and calibrate the model and results,so as to achieve the corresponding accuracy and availability level.In this paper,the above research results are sorted out and summarized,and the future prospects and plans for the research in this field are put forward.