IoT Device Positioning via Robot Cleaner and AI

The patent introduces a robot cleaner as an intermediary device that moves throughout the home environment to detect and identify IoT devices. The robot cleaner captures images of devices, extracts features, and transmits this information to the server, which then provides position information to the user interface. This intermediary approach solves the problem of obtaining device positions without requiring complex infrastructure changes to existing IoT devices.

The patent replaces traditional mechanical or manual methods of device identification with AI-based image recognition and processing. Instead of requiring users to physically locate devices or use complex scanning mechanisms, the system uses the robot cleaner's camera to capture images and an AI model to automatically identify devices based on visual features, substituting mechanical detection with intelligent image processing.

The patent pre-trains an AI model (such as a neural network) offline with extensive device images and features before deployment. This preliminary training allows the model to be stored in the server's memory in a ready-to-use state. When the robot cleaner captures images during operation, the pre-trained model can immediately process them without requiring time-consuming training during actual device identification, thus achieving both high accuracy and fast processing.

The patent creates a digital copy or representation of physical IoT devices through their images captured by the robot cleaner. Instead of physically examining or interacting with each device, the system creates visual copies (images) and processes these digital representations through the AI model. This copying approach enables accurate identification while saving time, as processing images is much faster than physical device inspection or manual identification methods.