A Pan Tilt Camera is mounted on the humanoid robot. In order to perform accurate self-localization, accurate camera calibration is necessary. As shown in the figure above, some white dots on the soccer field are used as targets during the calibration process.
Traditional camera calibration methods only estimate intrinsic parameters of camera. For our humanoid robot attending RoboCupSoccer Humanoid League, its camera is mounted on a pan-tilt unit. Both camera and pan-tilt unit are not high quality. Large localization error would be obtained on this type of robot because of the unavailability of extrinsic parameters and the inevitable error during the execution of the pan-tilt unit. In this project, we present an optimization based pan tilt camera calibration method, taking the execution errors of the pan-tilt unit into account. The calibration method is formulated as an optimization problem, which simultaneously estimates intrinsic parameters, extrinsic parameters and the execution error. Good localization accuracy is guaranteed because of the design of loss target. The effectiveness of the proposed method has been validated on our humanoid robot Wukong.
"An Optimization Based Vision Calibration Method for PTZ Camera's Errors in Model and Execution"
Dai, Qiqin, Rong Xiong, and Shen Li.
Procedia Engineering 15 (2011): 585-593.