Self-enhancing Featureless Robot Localization with Fixed Memory Tags

Abstract

This chapter presents experimental results on localization of indoor mobile robots with limited sensor capabilities. On the basis of a bayesian approach with probabilistic global occupancy knowledge of the environment the robot is able to locate itself near obstacles. Far from obstacles the estimate gets inaccurate due to slippage. To locally increase the accuracy of the estimation in these regions we suggest integrating memory tags without initial knowledge on their position. The memory tags could also be RFID tags or fixed sensor nodes of a stationary sensor network. These local storages are used to save a processed actual estimate of the robot when it is within communication range. Hence the tags passively get an estimate of their own position which in turn is used by passing robots. Experiments show that even without initial position information for robot and local memory tags and despite vague and partially erroneous information of the robot, the position information of the tags converges towards a good stationary value. Even wrong initializations of the tag positions are tolerated and corrected. Thus positioning of any robot in communication range can be enhanced.

Based on Self-Enhancing Robot Localization via Local Memory Tags, by Felix Hackbarth which appeared in the Proceedings of the 5th International Conference on Automation, Robotics and Applications (ICARA 2011).  © 2011 IEEE.