A self-localization estimation method with cooperating small robots Using camera positioning and movable landmarks

Abstract

Recently, there has been extensive research on robot control using self-position estimation. Simultaneous Localization And Mapping (SLAM) is one approach to self-positioning estimation. In SLAM, robots use both autonomous position information from internal sensors and data from external landmarks. SLAM can improve the accuracy of position estimation with a large number of landmarks, but it involves a degree of uncertainty and has a high computational cost, because it requires detection and recognition of landmarks through image processing. To overcome this problem, we propose a new method involving the creation of maps and the measurement of position using two cooperating robots that serve as moving landmarks for each other. This makes it possible to solve problems of uncertainty and computational cost with two-dimensional markers, because a robot needs to find only a simple two-dimensional marker, rather than feature-points landmarks. In the proposed method, the robots have a two-dimensional marker of known shape and size and have a camera kept at the front of the robots sensing the markers to determine distance. The robots use this information to estimate each other's positions and to control movement. To test the method experimentally, we used two real robots in an indoor environment. The result of the experiment revealed that the distance measurement and control error could be reduced to less than 3%.