Development Of Intelligent Navigation For A New Terrain Adaptable Wheeled Mobile Robot Based On ROS

In recent deacades,the technology of robots is fast developing.Different kinds of robots are used in different fields,which improve the productive efficiency and replace people in dangerous environment.The traditional mobile robots like the EOD robots,Fire robots etc.,have some defects in walking structure,operation mode,visualization,modularity.This paper investigates these defects of the traditional mobile robots,then designs a new walking structure and intelligent autonomous navigation experiment based on ROS.It realizes the functions of walking fast,obstacle surmounting combined with autonomous navigation.It has lots of research meanings and actual values.This paper investigates the wheel structure and track structure,and designs a new eight-wheel mobile platform.It has four groups of gears.With two degrees of freedom,each group has two wheels driven by the same motor.The new mobile platform can walk fast like the wheel structure and surmount obstacle like the track structure.Not like the track structure,the dynamical stability of obstacle surmounting is good.These features are convenient for the development of autonomous navigation.The traditional SLAM technology for navigation always avoids obstacles when facing them without any distinction.This defect has been improved in this paper.The installation height of the laser radar is related to the obstacle-surmounting ability.When the obstacles are lower than the installation height,they are ignored by the laser radar.Meanwhile the robot is capable of surmounting the obstacles.Thus different heights of obstacles are distinguished.But there still exists a defect with distinguishing different heights of obstacles during navigation.The robot can't judge whether slopes are surmountable or not,for all the slopes that are detected by laser radar will be regarded as unsurmountable.Obviously,this is not always true since the slope can be gradual.Depth camera is used to detect 3D data of the environment.Then the gradient is computed to decide whether to climb the slope or not.