Research On Method Of Outdoor Tracking For Quadruped Robots Based On Multi-line Lidar

In recent years,legged foot robot,as a mobile robot with special structure,has made remarkable achievements in scientific research and practical application.Its application is not only confined to people's life scenes,but also includes military applications and scientific research fields.Taking the large quadruped robot of the research platform in this paper as an example,due to its large size,strong load bearing capacity and the characteristics of foot structure,the main application scenario is the field combat environment,which can replace personnel to carry out combat tasks such as material delivery,battlefield mine clearance and fire strike.With the rapid development of artificial intelligence,the robot from the early years of human control movement to the present has basically been able to move autonomously to achieve a qualitative leap.The complexity of the use scene also puts forward new and higher requirements for the environment perception and autonomous behavior ability of the robot.In view of the field battle scenes,is often accompanied by field complex road conditions,obstacles more complex environment,experience requirements for personnel control operation robot is very high,so in this scenario if can let a quadruped walking robot to follow a pilot,can greatly reduce the experience requirements for personnel control and robot real-time autonomous path planning need to cost.In view of the above background,this paper mainly studies the following aspects:(1)the research status of mobile robots and sensors on target recognition and tracking at home and abroad is analyzed,and the system design of autonomous navigation and environment awareness of four-legged robots is analyzed to determine the data detection process and processing method of the system;(2)the research platform of this project is Scalf robot developed by the robotics center of shandong university.Due to the consideration of the application environment and scientific research cost,Velodyne VLP16 lidar is used as the point cloud acquisition equipment for equipment selection,and the data processing of point cloud is studied in detail on this basis.(3)the common point cloud segmentation algorithm is analyzed.Since the application scenarios of four-legged robots are usually not high-speed moving,it is studied and improved in the aspects of map construction and ground point cloud segmentation on the premise of meeting the real-time requirements of data processing.(4)in the aspect of target recognition,the improved conditional clustering algorithm and the enhanced reflection feature are adopted to identify the navigator;In the tracking aspect,kalman filter is used to ensure that the target navigator can not be lost after being identified,so as to ensure that the quadruped robot can follow and walk smoothly.