| With the development of artificial intelligence technology,more and more artificial intelligence technologies are applied to service robots,especially smart home service robots.But the technology level of these technologies is still a long way from realizing the real "robot housekeeper".As an important part of mobile robot,robot navigation technology has made great achievements in recent years.However,as the smart home robot is a system engineering,the existing robot navigation technology cannot be directly transplanted to the smart home robot.Firstly,the highly dynamic home environment will bring severe test to the existing robot navigation technology.Secondly,the existing robot navigation technology lacks the deep integration of human-computer interaction and grasping functions of smart home system.Therefore,this paper starts with the multisensor sensing technology of mobile robot,and on the basis of enhancing the robot’s environment perception,constructs a multi-level map to realize the navigation in highly dynamic environment.Secondly,this paper will also be deeply integrated with robot functions such as manipulator grasping and human-computer interaction to realize a whole set of perception movement grasping process.The main contents of this paper are as followsThe multi-sensor fusion technology of robot can realize the multi-modal perception of the environment.In this paper,two single line lidars and depth cameras are fused.The QR optical code reader placed in the center of the robot chassis is used as a reference.The sensor scanning features are used to establish pose constraints and optimize the relative pose between the robot and the sensor.The TF mechanism in ROS is used to manage the time and space information of multi-sensor.Combined with the existing indoor two-dimensional grid map building technology,this paper increases the spatial information of two-dimensional grid map through threedimensional reconstruction.At the same time,the dynamic obstacles are identified by the sensors,and the dynamic map is introduced.In addition,combined with the actual humancomputer interaction requirements of smart home robot,this paper introduces navigation semantic map and object semantic map for higher level robot control.In the navigation of smart home robot,in order to improve the pose estimation ability of the robot in the process of movement and make it have high precision positioning,this paper combines laser pose estimation with wheel odometer to realize the pose estimation module of mobile robot with high frame rate and high precision.Combined with the existing mobile robot navigation algorithm,through the motion control method based on PID,the robot terminal high-precision navigation is realized,which paves the way for the realization of robot stable grasping.Finally,the precision experiment is carried out on the robot platform,and the positioning accuracy is less than 0.02 m. |
