Visual And Ultrasonic Sensor-based Automatic Guided Vehicle Environment Perception

The thesis presents research on such key technique as architecture of Automated Guided Vehicle (AGV), environment perceiving based on multi-sensor information fusion and so on, with aims at improving the intelligence of AGV.The thesis summarizes the actuality of AGV. On the basis of analyzing architecture of mobile robot, the self-determination AGV architecture based on CCD and ultrasonic sensors is put forward, which can add or reduce some function according to the requirement of different user with good currency. Then the structure of QDU-I AGV, the choice and the layout of sensors, the hardware and software are studied. A least-square method is used to fit the calibration of the single ultrasonic sensor and a more accurate math model can be gained. A circum-circle method is further used to fuse the collection data, which reduce the fusion time and simplify the.calculation. A camera calibration method based on polynomial linearity fit for distorted picture is used and proved by the experiment, which brings higher camera measurement accuracy. The method for detecting object brim is discussed and used to found the geometry expression between the object and the ground. The environment perceiving experiment is carried and proves that the application of the multi -sensor system can advance the ability of AGV environment perceiving.