Lang, HaoxiangLi, Yunze2020-10-202022-03-292020-10-202022-03-292020-04-01https://hdl.handle.net/10155/1170This thesis describes the design and development of the novel autonomous UAV that has powerful GPU unit and potential sensing capability. It can run ROS (Robot Operating System) and compatible with packages developed under ROS. With the waterproof enclosure, it can land on the surface of water to do measurement of water parameters. Meanwhile, a human machine interaction mechanism using gesture recognition is developed and implemented for the developed drone to understand human commands. A novel way of image processing using depth information is developed for distance measurement. Meanwhile, a standard calibration method is applied and a vision-based object identification is developed and implemented in the on-board computer. It mainly achieves the function of licence plate identification and recognition. It includes a deep natural network and an OCR modular to achieve the identification, segmentation and recognition. The results showed the achievement of 95% success rate.enDroneDesign and developmentMachine visionMachine learningThesis