Research And Design Of Autonomously Following Robots

With the rapid development of artificial intelligence,service robots have received more and more attention,people's demand for service robots is increasing,and autonomous following robots are becoming an important part of service robots.The autonomous following robot designed in this paper based on ROS system can realize the functions of target following,simultaneous localization and mapping(SLAM)and autonomous navigation.The research content of this article is mainly divided into the following three aspects:1.This parts studies the SLAM technology based on lidar.First of all,the principle of grid map occupancy is analyzed,and the two processes of map acquisition and scan matching of the Hector algorithm are theoretically derived to construct the indoor environment map.Secondly,the Gmapping algorithm based on particle filtering is studied,and the steps of pose estimation are analyzed in detail.The recommended distribution function combining odometer information and lidar observation information is derived,and the map is constructed in the same indoor environment.Finally,the maps constructed by the two algorithms are compared and analyzed.It is found that the map constructed by the Gmapping algorithm added with odometer information is more complete and accurate,and the Gmapping algorithm is selected for map construction to prepare for subsequent path planning.2.Researched the path planning technology of following robots,and improved the dynamic window algorithm.In terms of global path planning,the A*algorithm and Dijkstra algorithm are deduced in principle.Compare experiments in the same environment,and select the A*algorithm with higher efficiency.In terms of local path planning,the evaluation function of the dynamic window algorithm(DWA)is improved.Discard the parameters with similar functions in the original function,and add parameters Angle and Heading to evaluate the angle relationship between the current path and the global path and the target point.The experiment compared the DWA algorithm before and after the improvement.Finally,set targets and obstacles in the constructed map,and use Rviz to observe the robot path planning in real time.The effectiveness of the A*algorithm and the improved DWA algorithm are verified,and the path planning function of the robot is realized.3.Researched the target following technology,and proposed a multi-information fusion following algorithm.First,the depth information and Camshift algorithm are fused to solve the problem of similar color interference.Use depth threshold to segment target and background.The depth information is merged into the original probability density distribution map.Through the depth difference between the first two frames of images,two different depths of interest are obtained and given different weights.This algorithm solves the problem that the robot cannot accurately follow when the similar colors interfere.Secondly,Kalman information is added to the Camshift algorithm fused with depth information.Follow through steps such as location prediction,search matching,and revision updates.This algorithm narrows the search range of the Camshift algorithm and solves the problem of loss caused by the target being blocked.Finally,based on the robot platform,the experimental comparison of the Camshift algorithm before and after the improvement is carried out.The multi-information fusion following algorithm finally achieved the expected effect,which confirmed the rationality and effectiveness of the method in this paper.