Research On Indoor AGV Positioning And Navigation Based On Multi-sensor Fusion

The paper addresses the problems of low positioning accuracy,sensor signal loss and large navigation track errors that AGVs are prone to in logistics and transportation.This paper carries out research on indoor AGV drive systems and positioning and navigation systems.In indoor sorting and transportation,where there are many goods and multiple AGVs operating simultaneously in the same environment,traditional magnetic navigation AGVs are not suitable.In this paper,a multisensor fusion-based indoor AGV is designed,which uses two sets of Mc Namee wheels to provide omnidirectional movement.Firstly,the drive unit of the indoor AGV is designed and the body configuration is designed according to the dimensional requirements.Then,the kinematic model of the AGV was carried out and an inverse/positive kinematic model of the AGV was derived.Finally,the software and hardware system of the indoor AGV was developed to lay the foundation for the subsequent multi-sensor fusion research.AGVs are prone to large position deviations and inaccurate positioning when navigating indoors.To address this problem,a new multi-sensor fusion scheme is proposed in this paper.Firstly,the environment and requirements for the use of AGVs are analysed,and suitable sensors are selected.Then,sensor fusion is investigated and a fusion positioning framework based on wheeled odometers,inertial measurement units,ultrawideband positioning systems and Li DAR is proposed.Research on extended Kalman filtering and adaptive Monte Carlo algorithms is carried out before implementing the framework.Finally,a mathematical model of the sensors is brought into the algorithm,leading to the design of an autonomous positioning and navigation algorithm for indoor AGVs.The adaptive Monte Carlo algorithm was also improved to address the problem of difficult particle initialisation.In order to run the AGV autonomous positioning and navigation algorithm better,an upper computer software was developed on the ROS system.The indoor AGV positioning accuracy and system robustness were tested by building a simulation environment,and the positioning error map was plotted using Matlab’s own tag tool.The experiments show that the indoor positioning based on multi-sensor fusion has a greater improvement in accuracy compared with single-sensor positioning.Finally,the feasibility of the improved adaptive Monte Carlo algorithm is verified.