This work proposes a 3D Printed Prosthetic Arm using Eye-TrackingArtificial Intelligence Visual Recognition Technology, composed of an AI vision device, AI box,AI prosthesis. The AI vision device provides a first-person viewthe point position through gazing. The AI box uses deep learning object detection technology to infer the type of object at the gaze point in the first-person viewcontrols the AI prosthesis to change gestures according to the object type to assist amputees to perform appropriate hand movements, including grasp, pick, lift, pointclick. The model in the AI box is obtained by the modified YOLOv4. This model can infer 10 typesa total of 25 objects with the accuracy up to 99.

This work is different from existing electronic prosthetic products, such as the unintuitive operationslow action response. It is the first to use eye tracking combined with artificial intelligence edge computing object recognition technology to identify the object at the user＇s first point of view on the screen,real-time Control the 3D printed prosthetic arm to perform appropriate hand movements, to achieve "what you see is what you get" intuitive operation modesensitive response speed. At present, it can identify 25 kinds of common living objects in 10 categories, with an average accuracy rate of 99.

As the demand for prosthetics increases year by year, the market size also grows year by year, but it faces challenges such as high cost, ease of usecomfort. After winning the 2021 FITI Outstanding Entrepreneurship Award, this work established a new company, Yizhi Technology, for commercial development,will cooperate with domestic assistive device manufacturers to provide customized productionsustainable services. As for the eye tracking technology mastered, it can be extended to other fields of virtualreal integration operation applications in the future.