The Research On Robot Autonomous Motion Planning Based On Multiple Sensors

In the study of contemporary robots,autonomy is a very important indicator,especially in the unknown environment.The robot can construct the distribution map of obstacle and path planning by itself through the autonomous perception surrounding environment,then,finished task as soon as possible.This has great significance for the research of robot and the development of artificial intelligence.But in research and applications,the positioning accuracy is not very high when robot is in the unknown environment,as a result,unknown environment maps are built inaccurately.At the same time,the path planning algorithm under unknown environment is complex and difficult to implement.Therefore,this paper will mainly study in three parts include robot localization in unknown environment,the construction of environment map and path planning algorithm analysis in unknown environment.It is important to build a suitable robot hardware platform in order to collect data and verify the results better.This paper adopted a two-wheel four-drive robot chassis which can independent control and in situ circular motion.The robot use gyroscope to record mileage and odometer to record heading angle,it can also avoid obstacles through information from ultrasonic sensor and detected surrounding obstacles through laser sensor.We should use coordinate transformation to transform local coordinate from robot’s view and polar coordinate from laser sensor into global Cartesians coordinate system,because the robot is always scan obstructions self-centered when it moved.So,coordinate transformation is conducive to the construction and analysis of the map.The error of odometer and gyroscope will accumulate with the distance increase,it may affect the robot path planning accuracy.This error is usually corrected by external sensor information,such as Bluetooth road or infrared road signs.But this paper mainly study in unknown environment,where will be no artificial pre-set road signs.Therefore,this paper presents an error correction method based on obstacles.The robot scan obstructions through laser when it is just start moving,then use the least squares method to extract the contour and draw it in raster map as a reference,after that the robot fused data from sensor and reference to reduce the sensor error when it is continue moving.In order to describe the obstacle information better,this paper fit straight line contour and curve profile respectively,but in multi-obstacle environment,the SEF algorithm is used to divide the different obstacles before fit.In this paper,the human obstacle avoidance behavior is analogous to the movement of the robot.The motion behavior of robot is divided into child behavior such as approximate target,in situ turn and contour tracking which is easy to implement.Then the traditional Bug algorithm is improved,update the robot and the target point of the connection equation constantly,the motion behavior of the robot is added to the improved algorithm to finish simulation and comparison.