Researching On Mobile Robot Autonomous Perception Of Environment Method Based On Visual Information Fusion

As the application in the field of mobile robot is gradually expanded,its working environment is also becoming complicated.Mobile robot autonomous perception system is responsible for the environment of robot state of internal and external environment information,and is the basic of the robot decision.The proper use of the sensor will directly affect the mobile robot that can accurate perception of the environment information,but for a lot of the complicated environment and changeable situation,such as rough,obstacles appear frequently.Only relying on a single sensor can hardly meet the requirements.Therefore,based on CCD binocular vision camera and supplemented by ultrasonic sensors and a laser range finder information fusion technology in environment perception can be more accurate and efficient.This paper first has an overview on the environment of autonomous mobile robot perception of domestic and foreign research situation at present,and the research contents and key technologies involved are summarized.Secondly,the environment of mobile robot autonomous perception theory were in-depth studied,mainly including mobile robot obstacle avoidance method and the three-dimensional physical environment scene reconstruction,even based on visual method of information fusion theory.In this paper,the automatic obstacle avoidance for mobile robot and what the road can be passed will be intensively studied.Again,in view of the environment of the mobile robot autonomous perception method set up the experimental platform.Software and hardware platform is divided into two parts.Software mainly includes the MATLAB and VC++.MATLAB can fast realize the function of image processing,facilitate to verify this algorithm and the simulation analysis,and VC++ can be developed with good man-machine interface of user software.Hardware based on the unmanned vehicle UGV-1 platform,to integrate the CCD binocular camera,ultrasonic sensors and a laser range finder,get all kinds of sensor fusion data.All algorithms is validated with the real physical environment information in validity and reliability.Finally,put forward with the combination of image texture features and color features to identify terrain,and identify the terrain through evaluation by an experiment.The simulation experiment can be divided into four processes: the road imageacquisition,road image processing,analysis of trafficability characteristics and simulation movement.By summing up analysis,verify the validity and enforceability of evaluation algorithm.