Accurate Visual Loop-Closure Detection Using Bag-of-Words for Multiple Robots
Jung H. Oh, Seung-Hwan Lee, and Beom H. Lee
Department of Electrical and Computer Engineering, Seoul National University, Seoul, Republic of Korea
Abstract— We propose a method to detect loop-closures in a simultaneous localization and mapping (SLAM) problem for multiple robots. Each robot should be able to detect other robots’ previously visited locations from camera measurements. To identify these places, our approach adapts the bag-of-words method in image recognition, and improves it by applying a Gaussian filter and a logistic function to correct the similarity scores. We can detect the robust loop-closures using only visual information of multiple robots. Experiments are performed to verify the effectiveness of the proposed method in indoor environments.
Index Terms— loop-closure, mobile robots, SLAM, visual feature, bag-of-word
Cite: Jung H. Oh, Seung-Hwan Lee, and Beom H. Lee, "Accurate Visual Loop-Closure Detection Using Bag-of-Words for Multiple Robots," Jounal of Automation and Control Engineering, Vol. 3, No. 5, pp. 354-359, October, 2015. doi: 10.12720/joace.3.5.354-359
Index Terms— loop-closure, mobile robots, SLAM, visual feature, bag-of-word
Cite: Jung H. Oh, Seung-Hwan Lee, and Beom H. Lee, "Accurate Visual Loop-Closure Detection Using Bag-of-Words for Multiple Robots," Jounal of Automation and Control Engineering, Vol. 3, No. 5, pp. 354-359, October, 2015. doi: 10.12720/joace.3.5.354-359