Research On Fusing Autonomous Navigation Information In The Calibration Of Distributed Metrology System

Large-scale precision metrology has become a key support for advanced manufacturing and assembly.The traditional single-instrument system has difficulties meeting the requirements of parallel measurement and multi-target tracking in those scenarios.In recent years,the Distributed Metrology System(DMS)has shown its advantages of automation and efficiency,which makes it an inevitable trend in intelligent manufacturing.Extrinsic calibration is a fundamental requirement to obtain the coordinate transformations from the world coordinate to the inter-related instrument coordinates,and a critical process before measuring.Taking the workshop Measuring and Positioning System(w MPS)as an example,however,locating and orienting a network of transmitters relies heavily on human operations and subjective experience,which makes the process complicated and tedious,contradicting its inherent advantages.Besides,changes of any measurement node will result in the need for re-calibration of the whole system.Therefore,the existing methods cannot meet the goal of long-term automatic maintenance of the DMS.Given the fact that Automated Guided Vehicles(AGVs)are used across industry sectors and many recent studies have been focusing on mobile robot navigation based on DMS,we would like to investigate an AGV-based(self-)calibration method of DMS that can work automatically without usual requisite manual operation.This thesis describes work on obtaining prior information(such as the transmitter distribution and environmental characteristics)using active SLAM techniques,and planning the calibration points(the poses of calibration receivers)based on the prior information.Experimental results demonstrate accurate and efficient calibration,which represents a step towards developing intelligent DMS that have the capacity of long-term operation without surveillance.The main contents are as follows:1.We improve the error-distribution locating and orienting algorithm using resection and adaptive Monte Carlo localization.The constraints are formulated from information gathered at planned calibration points by w MPS receivers mounted on the AGV.2.We use SLAM in conjunction with resection to estimate the transmitters' poses and build a map of the measurement environment.Then we develop a planning algorithm to determine the valid distribution of calibration points and calculate the best orientations of each calibration receiver based on the prior information.3.We implement an autonomous exploration and mapping algorithm on AGV in order to obtain the prior information of the unknown measurement field.AGV trades off between exploring new areas to complete the mapping and exploiting the visited areas to maintain good localization,based on an information-theoretic utility function.