Research On Autonomous Navigation Of Mobile Robots In Crowd Environment

With the development of artificial intelligence and robot technology,mobile robot assisted service industry upgrade is receiving widespread attention.For mobile robots to work properly,they need to have the ability to sense,locate,and navigate.Among them,autonomous navigation capability is a basic and important capability.In traditional scenarios,mobile robots only need to move in a static environment.However,in the service industry,mobile robots often need to move in a crowd environment,which will bring great challenges to the navigation of mobile robots.This paper mainly studies the autonomous navigation technology of mobile robots in a crowd environment.Based on the existing navigation algorithms and combined with the state of pedestrian movement,an improved navigation algorithm is proposed.The aim is to solve the problem of pedestrian motion when navigation and obstacle avoidance trajectories are not considered when existing mobile robots work in a crowd environment.The main work of this article is as follows:1.Research on Pedestrian Detection Algorithm Based on Lidar.Aiming at the small amount of information in single-line lidar,direct detection of pedestrians will cause more false detection problems.A static laser point cloud background removal algorithm based on known environmental maps is proposed to improve the practical application accuracy of single-line lidar pedestrian detection.On the basis of the existing description of pedestrian geometric features,statistical features are added to improve the detection accuracy of pedestrians when their legs are merged.2.Research on Multi-Pedestrian Tracking Algorithm Based on Kalman Filter.According to the characteristics of pedestrian movement,a Kalman filter tracker model is established,which can effectively track the movement state of pedestrians.Aiming at the problem that the existing Kalman filter is not robust enough to track multiple pedestrians,Euclidean distance and laser reflection intensity are used to match the pedestrians before and after the frame,which improves the robustness of multi-pedestrian tracking.3.Research on path planning algorithm in crowd environment.This paper proposes a high-dimensional time-domain dynamic cost map modeling method,which solves the problem that the existing navigation algorithms do not take into account the state of pedestrian movement and obtains a better path suitable for mobile robots to walk in a dynamic pedestrian environment.In the TEB local path planning algorithm,dynamic pedestrian constraints are added to solve the problem that existing local paths will compete with pedestrians forward when facing dynamic pedestrians,and a more friendly path for pedestrians can be obtained.4.Experimental research on autonomous navigation algorithms for mobile robots in crowd environment.Set up a mobile robot simulation environment and an actual mobile robot platform,design pedestrian detection,pedestrian tracking,and path planning verification experiments,and experimentally verify the performance and overall performance of each module of the system.