High voltage overhead transmission lines will be covered with a large amount of ice and snow in winter, and the changing weather conditions will change the shape and texture of the ice and snow, which in turn leads to notable disadvantages in automatic de-icing methods that rely on echo signals. Under the interference of stored echoes, there are obvious drawbacks. To this end, research is being conducted on the uninterrupted deicing technology of high-voltage overhead transmission lines based on small mobile robots. Employing a small robot equipped with binocular vision sensors, the system captures real-time video images of the transmission lines. It then extracts the texture features of appearance of the line from the video images and inputs these features into a neural network for the identification of ice accumulation conditions. Utilizing the principle of similar triangles, the three-dimensional spatial coordinates of the iced transmission line are derived. By using proportional-integral-derivativ control, the robot robotic arm can overcome obstacles and use a portable ice melting device to remove the ice layer covering the power line, completing uninterrupted ice melting operations. The results show that the appearance of the transmission line is fuzzy and irregular before ice melting, and after ice melting, the area of the transmission line became clear and visible, revealing the color and texture of the line itself. This indicates that the ice melting operation has achieved significant results and proves the effectiveness of the technology.