Research On Wheelchair Following Control System Based On Computer Vision

With the continuous development of mobile robots,as a key technology,target tracking technology has played an important role in promoting the development and research of mobile robots,so it has become one of the hotspots of many scholars.At the same time,with the acceleration of the aging of the population and the increasing attention of the society to people with disabilities such as the elderly and the disabled,wheelchairs as their auxiliary tools will greatly improve their self-care ability and develop Serve robots for carriers to assist them in restoring health and improving their quality of life with broad application prospects.This article is supported by the Xianyang Science and Technology ResearchProject "Research on Intelligent Wheelchair Control System with Target Automatic Following Function and Fall Detection Function"(2018K2-06),With wheelchair as the carrier,this paper focuses on the dynamic target tracking control technology with target detection and target tracking as the core.On this basis,an intelligent wheelchair system with automatic follow function is designed,which is based on the embedded hardware platform with multiple sensors.The system can detect and follow wheelchair users through computer vision sensors,and achieve safe and accurate obstacle avoidance in the process of following,so as to provide help for people in the rehabilitation stage or weak body.The main research contents are as follows:(1)System overall plan designAfter analyzing the task requirements of the wheelchair with target following function,the overall scheme design of the entire wheelchair following system was completed:that is,the video image of the target was obtained in real time through the computer vision sensor,and the target detection tracking algorithm and binocular vision positioning algorithm were used to obtain The position information of the target,combined with lidar ranging sensor,attitude sensor,GPS sensor and so on to obtain the wheelchair’s own position and surrounding obstacle information,it is carried into the embedded platform,so as to control the wheelchair motor’s movement state to achieve the goal of following and avoiding obstacles in the process of moving.(2)Resear-ch on target detection and tracking algorithmsThe target detection tracking and positioning algorithm is mainly divided into three parts.The first is the detection of the target.For the background subtraction of the inter-frame difference method and the Gaussian mixture model,there are problems such as incomplete target detection information and background false detection.The background subtraction method of Gaussian mixed background modeling,and the superiority of the algorithm is verified through simulation;the second is the target tracking algorithm,and the particle filtering algorithm and the TLD tracking algorithm use a single feature to perform target tracking with target drift and loss.A particle filter-based TLD tracking algorithm based on multi-feature fusion is proposed.At the same time,the fusion direction prediction algorithm improves the detection module and improves the processing speed of the detection module.Simulation experiments show that the algorithm can achieve long-term safe and stable tracking.Finally,the camera calibration,matching and binocular ranging and positioning experiments were completed through corresponding experiments,which laid the foundation for the realization of the target following control system.(3)Software and hardware design of wheelchair following systemAccording to the overall scheme and functional requirements of the system,the raspberry pie 4B processor based on arm cortex-a core is determined as the upper computer controller for video image data and lidar data processing,and the STM32 based on arm Cortex-M3 is determined as the lower computer controller,the power circuit,the controller circuit and the peripheral interface circuit were designed,processing of wheelchair control system,other sensors in the system were introduced and selected.In order to realize the real-time and accurate tracking and obstacle avoidance function of the wheelchair following system for the user,the main program design of the following wheelchair control system and the target tracking subroutine,the system following and obstacle avoidance control subroutine,and the attitude sensor data acquisition subroutine were completed.And the design of GPS positioning subroutine.(4)Functional testAfter the system is set up,follow the obstacle avoidance test in indoor static environment,outdoor environment,and multiple personnel interference.At the same time,test and analyze the GPS positioning effect to verify the effect of each function of the system.In this scenario,the wheelchair following system has a better follow-up obstacle avoidance effect for the target person.Through the test of various functions of the wheelchair following system,it shows that the system can not only complete the real-time follow to the user,but also realize the safe avoidance to the obstacles.The realization of the system can provide convenient services for people in the stage of rehabilitation or weak body,improve the users’ self-confidence in life,improve their quality of life.Moreover,the designed system has good versatility and transplantability,and It can be transplanted to other moving bodies,at the same time,the key technical problems in the research of automatic follow wheelchair also have a certain help to the research and development of mobile robot.