Research On The Problem Of Passing Through Narrow Door For Self-driving Wheelchair And Its Implementation

Since 2000,with the deepening of China’s aging,the number of people with disabilities and the elderly with impaired mobility is rising rapidly,the demand for intelligent wheelchairs is becoming urgently more and more.For many wheelchair users,their physiological response speed is slow,and the ability to control a wheelchair is also weak,that making it difficult for them to use the traditional non-intelligence wheelchair safely and conveniently.This thesis takes smart wheelchairs as the research topic,mainly researches the problem of wheelchairs passing narrow doors under the condition of self-driving,and looking for safe,comfortable and efficient methods.Compared with outdoor road navigation,there are problems such as the inability to obtain accurate GPS positioning information in the near-door environment and the high positioning accuracy requirements in the process of self-driving wheelchair passing narrow doors.At the same time,there are many kinds of indoor doors,which are difficult to predict,and the control of wheelchair will be disturbed by the load change and other factors.Aiming at the above problems,this thesis mainly researches from the three aspects of the design of wheelchair software and hardware system,indoor positioning and indoor navigation.The hardware system of this paper is divided into four parts: remote control module,main control module,motor power module and multi-sensor module.The remote control module is responsible for human-machine interaction;the main control module is responsible for the overall control of system resources and the interaction with the embedded operating system;the motor power module is responsible for the control of wheelchair linear or differential motion;the multi-sensor module is responsible for environmental perception and obtaining the wheelchair posture.The software system is mainly divided into local main control program,remote main control program and sensor control program.The local main control program is responsible for providing positioning and navigation for the wheelchair;the remote main control program is responsible for controlling,calibrating and alarming the wheelchair through the mobile phone;the sensor control program is responsible for collecting,calibrating and correcting the sensor data.For wheelchair indoor positioning,this paper uses the method of INS positioning based on EKF and UWB positioning based on least-square,and uses the UWB/INS combination positioning based on cascaded EKF to achieve high-precision positioning in the indoor environment.The method first solves the INS positioning data by EKF quaternion,then uses the results to determine the UWB positioning NLOS error,and finally uses the EKF-based UWB data to calibrate the INS data to improve the accuracy of indoor positioning.For indoor wheelchair navigation,this thesis uses a multi-sensor fusion method to achieve the perception of the environment.The solid-state lidar installed in the front of the wheelchair is used to obtain the point cloud data in front of the wheelchair.EKF is used for data processing,and the data obtained by wheelchair attitude sensor is fused to obtain environmental information.Then the results are used for path planning and automatic obstacle avoidance.This paper improves the traditional artificial potential field method,avoids the shortcomings of easy entry into local minimum traps,and uses RS to optimize the planning path,and realizes high-quality path planning and automatic obstacle avoidance through virtual gravity and repulsion.The self-driving wheelchair designed and implemented in this thesis has carried out multiattitude passing narrow doors experiments in the simulation environment and indoor real environment.The results of multiple experiments show that it can meet the requirements of safe,comfortable and efficient passing through narrow door for self-driving.