Research On Real-time Position And MAP Creation Of Mobile Robot Based On RGB-D Sensor

With the continuous development and progress of the national economy and science and technology,in the needs of industries such as industry,agriculture and military,the research of autonomous mobile robots cannot be delayed.The emergence of real-time positioning and map construction technology has solved the problem of positioning and mapping of robots in an unknown environment.It is one of the top ten technologies in the field of robotics in the 21 st century.Whether it is the popular unmanned car,or the use of more and more sweeping robots in the home life,it is especially important that the robot itself can accurately locate and accurately capture the traversed trajectory during use.The common SLAM algorithm still has many problems in pose estimation,loop detection and backend optimization.Based on this situation,this paper proposes an improved algorithm for real-time position and map creation of mobile robots based on RGB-D sensors,which improves the accuracy of robot position and reduces the robot's failure to encounter frame loss caused by sudden problems during actual motion.First of all,in the traditional SLAM algorithm,when calculating the pose of the robot,the matching pose estimation between the current frame and the previous frame is used.When a current frame has a large pose error due to the influence of noise,the bit of the subsequent frame is caused.The pose estimation differs greatly from the pose of the actual frame,causing the mobile robot to fail to locate.In order to overcome this shortcoming,this paper proposes a method of introducing a matching graph,which consists of frame key points and key frames.The first frame acquired is set as the key frame,and the frame is initialized with the extracted frame key pair matching map.After calculating the current frame pose,the key points of the current frame are extracted,and the key points of the key points and the matching map are calculated by using a PnP(Perspective-n-Point)algorithm to calculate the pose of the current frame.And the matching map is continuously updated.The distance between the frame key points in the matching map and the current frame is continuously determined.If the distance is too far,the frame key points are deleted,and the frame key points of the current frame after the matching are added to the matching map.By introducing the matching graph method,the pose estimation accuracy is improved and the robustness of the SLAM algorithm is increased.In the back-end optimization,this paper uses pose optimization.Common graph optimization SLAM uses random detection or KD tree loop detection,and loop detection is inefficient.This paper adopts a loop detection algorithm based on K-tree dictionary.The time of loop detection is reduced,the detection efficiency is improved,and the accuracy of loop detection is improved.This paper makes full use of the loop detection algorithm based on K-tree dictionary to retrieve the advantages of high efficiency,and innovatively designs the frame-dropping loop detection algorithm,which reduces the risk of the frame-dropping failure caused by the sudden problem of the robot in the actual motion process.Finally,based on the Ubuntu platform,the designed SLAM algorithm was simulated.The simulation comparison experiments between the SLAM algorithm using matching graph and the SLAM algorithm without matching graph are carried out.It can be seen that the improved SLAM algorithm can construct the room map more accurately,which proves the superiority of the matching graph algorithm introduced in this paper.The SLAM algorithm using frame-dropping loop detection is compared with the SLAM algorithm without frame-dropping loop detection.By comparing the motion estimation trajectory and the real motion trajectory of the two,it is proved that the improved SLAM algorithm can realize the precise positioning function of the mobile robot,and at the same time verify the robustness of the SLAM algorithm using frame-dropping loop detection to the frame loss problem in the actual motion of mobile robots.Finally,the verification experiment is designed.By judging whether the root mean square error of the absolute trajectory error is at the centimeter level,the algorithm can realize the real-time positioning and map creation of the mobile robot,which proves the feasibility and effectiveness of the algorithm.