Pose Estimation And Autonomous Navigation Of Quadruped Robot

This paper mainly focuses on the pose estimation and map construction of autonomous mobile robots.With the development of science and technology,computer technology has brought great convenience to mankind.Intelligent and autonomous robots are gradually appearing in the sight of the public.For example,micro air vehicles,unmanned vehicles,Mars rover,etc.Pose estimation and the construction of the surrounding environment map are important parts of the robot.This paper uses a quadruped robot as a carrier,using binocular cameras,inertial measurement elements and lidar information for multi-sensor fusion to obtain robust,high-precision poses and coherent mapsThe ability to locate and build maps refers to the use of environmental observation information to estimate changes in one's pose and movement trajectory,and at the same time to build a map of the surrounding environment.In the early days,these two technologies mainly relied on satellite navigation systems,inertial measurement elements and wireless signals.Satellite navigation can only work outdoors,and the inertial measurement components have the shortcomings of accumulated errors,so that the early robot technology has not been greatly developed.Nowadays,positioning technology based on vision and lidar information has become a hot spot.The former's low cost and the latter's high-precision features are deeply loved by many companies.However,in the case of strenuous exercise alone,the visual information is easy to diverge and the map is sparse,so it cannot be used for navigation.The individual lidar information is also easy to diverge in an environment with plains and few buildings.These factors are fatal to outdoor mobile robots.In order to solve the above problems,this paper proposes a multi-sensor positioning scheme based on vision,lidar and inertial measurement elements and a compact and coherent navigation map representation.Traditional robots are mostly wheeled,which makes it difficult to complete tasks when the road is bumpy.Adopt foot structure,can face any road conditions.Highprecision,high-robust positioning and coherence,and compact map representation also provide a basic guarantee for quadruped robot navigation,enabling rescue missions,unmanned transportation,and planetary exploration missions to be implemented.