Global Path Planning Algorithm And Experimental Research Of Wheeled Intelligent Vehicle

Against the background of mobile intelligent car's indoor navigation,the mobile intelligent car's host computer software and hardware development,global path planning,place recognition and environmental modeling and other issues are deeply researched.Firstly,the hardware platform and software platform of mobile intelligent vehicle are built and configured.In hardware aspect,mobile robot Pioneer3-AT equipped with SICK LMS-200 laser radar sensor,vision sensor Kinect and ultrasonic sensor,in addition to the embedded board BeagleBone Black as the host computer of mobile robot is studied.In software aspect,Aria software,MobileSim software,Ubuntu system and robot operation system are migrated to the embedded board BeagleBone Black,build an experimental software platform in the notebook and the BBB board,and establish a distributed system based on ROS.In second,in path planning algorithm aspect,this paper introduces the principle and application of A star algorithm,do simulation experiment on the MATLAB,do the simulation experiments of the robot in the simulation software MobileSim,and do a lot of experiments on the P3-AT experimental platform to verify the effectiveness of A star algorithm,and summarizes the advantages and disadvantages of A star algorithm in indoor path planning practice.Thirdly,this paper studies the problem of room identification based on laser data.Because of its high precision,the laser distance sensor is one of the most commonly used sensors for the mobile robot.This paper use the circle projection and extreme learning machine algorithm to classify the data samples,use SICK LMS-200 laser radar to collect sample data of 9 indoor places and do experiments,the average recognition accuracy rate is about 74%,and the extreme learning machine algorithm performs better than support vector machine.In order to further verify the effectiveness of the algorithm,we also do experiments on common data sets,the test average recognition accuracy rate is about 70%,the results show that our recognition method also have a good effect in public data sets.A series of experiments show that our recognition method is effective,and could improve the mobile robot's obstacle avoidance and navigation performance combined with global path planning algorithm.Fourthly,we do some experimental research on the environmental modeling problem of mobile robot,using laser radar and Kinect sensors to collect experimental data and do environmental modeling experiments in different indoor rooms,mainly including laboratory,corridor,elevator,fitness room,conference room,basement,kitchen and bathroom.Finally,the obtained results are summarized and future work is addressed.