Research on self-location estimation applicable to various environments

- Self-localization in an environment with obstacles not on the reference map -

Abstract

AGV was run by integrating multiple self-location results in one environment. The feature of self-position integration is that there is no problem if it cannot be estimated by one method but can be estimated by another method. There are three position estimation methods used for integration: Particle Filter, ICP Matching, and SLAM. SLAM is performed by ICP Matching. The method of integration is weighted averaging and the parameters are fixed. In the integration, when the estimation cannot be performed by a method other than SLAM, the parameter is set to 0 and the result is not used. SLAM does not set the parameter to 0 because it is less affected by the environment. As a result, the self-location integration did not break down and the AGV was able to run.