A Localization Method Based on Large Scene Tracking and Target Details Observation About Small Quadruped Robot Under Global Vision

摘要

This paper resolved the problem of observing the target's high definition details when the global camera observing a large range field, in the process of mobile robot's location and navigation under global vision in outdoor environment. Two cameras with different focal lengths were used. The fisheye camera with short focal lens was used for the observation of large scenes, and the camera with long focal lens fixed on a PAN/TILT Head was used for observing the target's details by amplifying it. In this paper, the Particle Filter Algorithm with the weighted RGB color histogram was used for tracking target in the image caught by the fish-eye camera. The OcamCalib Tool was used for calibrating the fish-eye camera in MATLAB, to establish a mapping model from the image coordinate system to the lens coordinate system, calculating the vector of the target in the lens coordinate system. This vector were used for calculating the rotation angle of two joint of the PAN/TILT Head. The rotation speed of the electromechanical actuator were controlled indirectly by three layers of loop, so that the PAN/TILT Head was rotated smoothly to the reference angle position. Using a long focal camera to amplify and observe the target, the QR code attached to the mobile robot was observed in high definition. Extracting the nested contours to calculate the direction of the QR code and the orientation of the robot. In the end, the effectiveness of the method is verified by experiments.