| Quadruped robots have great application prospects in outdoor tasks such as outdoor exploration and industrial inspection because of their multiple advantages such as flexible movement and strong terrain adaptability.In order to accomplish the tasks in these complex environments,quadruped robots are inseparable from environment sensing and autonomous navigation technologies.Sensing real-time environmental information and constructing a highly accurate and navigable map for path planning and motion decision deployment are necessary to realize the outdoor navigation function of quadruped robots.In this paper,we focus on the construction of the perception system,the construction of the navigation map and the implementation of the autonomous navigation method for the operational requirements of outdoor navigation movement of quadruped robots:1.Build a quadruped robot environment sensing system.Firstly,we analyze the demand of outdoor environment perception of quadruped robot,and reasonably select sensors such as LIDAR,camera and inertial measurement unit as perception components.Then,we study the calibration method of sensor external reference based on point cloud alignment,and calibrate the position relationship between sensors to obtain the spatial transformation matrix.Finally,a communication connection is established between the remote control side,the motion controller side and the perception side to complete the construction of the environment perception system of the quadruped robot and provide the hardware basis for the research of environment perception and autonomous navigation.2.Investigate a tightly coupled lidar inertial odometry with closed-loop detection function.Firstly,the lidar point cloud is feature extracted to establish key frames,and the pre-integrated IMU data is used to correct the point cloud distortion of key frames.Secondly,the pre-integrated IMU data is used as pre-integrated factor constraints,the lidar key frames are matched with the subgraph formed by multiple frames,and the obtained poses are used as lidar odometry factor constraints,and the results of Scan Context-based method detection are used as the closed-loop factor constraint.Finally,the multiple factors are optimized for the factor map to eliminate the global accumulation error and provide real-time positioning information and global optimized poses for the environmental mapping study.3.Construct a global two-dimensional grid navigation map based on correlated elevation information.Firstly,the lidar point cloud is mapped in the grid map for height estimation by combining the bit poses provided by the odometer.Secondly,a ray-tracing based dynamic obstacle processing method is investigated to remove the wrong height estimation left by dynamic obstacles.Then,the slope and smoothness of the ground are calculated based on the height information localized in the grid to obtain the passability of each grid.Finally,the passability score within each grid is compared with the threshold score to determine whether the grid is idle,occupied or unknown,and finally a global navigation cost map is generated to lay the foundation for outdoor navigation research of quadruped robots.4.Investigate a navigation control method applicable to outdoor motion.First,a quadruped robot localization method based on normal distribution transformation is investigated to obtain real-time positional information through lidar point cloud and point cloud map alignment.Then,based on the global cost map,the low-order global route is generated by the A*-based global path planning method,and the real-time control command of the quadruped robot is calculated by the D WA-based local path planning method with velocity sampling and trajectory evaluation to realize dynamic obstacle avoidance.Finally,the control commands are sent to the motion control terminal in real time to realize the effective combination of perception and motion control and complete the two-dimensional navigation of the quadruped robot based on 3D positioning in the outdoor environment. |
PDF Full Text Request