國立臺灣師範大學電機工程學系Chen-Chien HsuChing-Chang WongHung-Chih TengCheng-Yao Ho2014-10-302014-10-302012-06-010253-3839http://rportal.lib.ntnu.edu.tw/handle/20.500.12235/32122A self-localization method entitled Dual-Circle Self-Localization (DCSL) for mobile robots in a soccer robot game filed is proposed in this paper. To evaluate the performance of the proposed localization method, an omnidirectional vision device is installed on top of the robot to capture images for recognizing objects in the environment via a hybrid scanning method. Three reliable objects recognized are then chosen as landmarks to calculate the position of the robot in the game field via the proposed DCSL based on relative positions between the robot and landmarks in the omnidirectional images. Because simple geometric manipulations are only required via the DCSL algorithm, localization efficiency can be significantly improved while maintaining a relatively smaller mean error in comparison to Monte Carlo Localization techniques as demonstrated in both simulation results and practical experiments in this paper.