Systems, Man, and Cybernetics

 

Home

• [P1007] Active-Semantic Localization with a Single Consumer-Grade Camera
  Author(s): Chuho Yi, Il Hong Suh, Gi Hyun Lim, Byung-Uk Choi
  
  

  Abstract
  In this paper we assume the problem of an active localization which requires massive computation. To solve this, our study proposes abstracted measurements that consist of qualitative metrics estimated by a single camera. These are contextual representations to be shared by humans and robots consisting of perceived landmarks and their spatial relations. Then, we develop the Markov localization method to support our contextual representations with which the location of a robot can be sufficiently estimated. In contrast to the passive approaches, our approach actively using greedy technique selects a robot's action to improve the localization results. The experiment carried out in an indoor environment indicates that the active-semantic localization leads to more efficient localization.

  
   

Full Text (PDF)