Research On Hotel Service Robot Algorithm Based On Visual SLAM Front-end And Path Plannin

Due to the increasing demand for manpower in the hotel industry in recent years,and avoiding unnecessary contact as much as possible during the epidemic,how to efficiently use hotel service robots to further improve the efficiency and quality of life in the hotel industry has become a hot research question.Visual odometry and path planning are two important parts to improve the overall performance of hotel service robots.The former is used as the front end of the visual SLAM,and it can usually estimate the robot’s motion trajectory based on the previously estimated camera pose transformation;the latter can find the optimal path of the robot with the least cost value and no collision with obstacles through the corresponding algorithm.Aiming at the characteristics of hotel indoor scenes,this paper first proposes a visual odometry method based on improved point and line feature fusion,which effectively improves the accuracy of the visual odometry system;secondly,a global path planning based on improved A* algorithm is proposed,This method can keep the path trajectory safer on the premise of keeping the pathfinding search efficiency higher.(1)In hotel premises,there is often a lot of low-texture environment information,such as walls,floors,etc.For visual odometry systems that perform pose estimation based on point features,the accuracy of the visual odometry system is usually reduced due to the lack of matching feature points in the hotel environment.This paper proposes a visual mileage calculation method based on improved point and line feature fusion.The method uses adaptive threshold-based fusion quadtree Oriented FAST algorithm for point feature extraction,while using GMS algorithm for feature fine matching,which is aimed at image frames The improved LSD line segment detection algorithm is used to extract line features,and through the nearest and second-close distance ratio method for fine matching,and finally the point-line feature weighted fusion method is adopted,so that the proportion of point features and line features can be reasonably allocated and improved The performance of the visual odometer system.Experiments show that compared with the ORB-SLAM2 algorithm,the absolute trajectory error and the relative pose error are reduced,which effectively improves the positioning accuracy of the visual odometer system.(2)In hotel premises,under the premise of ensuring more efficient services to guests,it is also necessary to ensure that service robots can safely reach their destinations.Traditional path planning algorithms often cannot guarantee the timeliness and safety of service robots during their work.Aiming at the shortcomings of poor real-time performance and insufficient security of traditional A* algorithm in global path planning,this paper proposes an improved A*algorithm that integrates JPS and safety weight square matrix to optimize the path of mobile robots.Simulation experiments are performed using grid maps of different sizes.The experiment proves that compared with the traditional A* algorithm,the improved A* algorithm proposed in this paper,which combines JPS and security weight square matrix,can maintain higher efficiency of pathfinding and search.It can ensure that the path trajectory is safer.