Moving-Robot Location Recognition Using Time-Based Multi-Map Selection

The patent divides the location recognition system into multiple maps, each corresponding to different illumination conditions or time periods. Instead of using a single map that degrades under varying conditions, the system segments the environment into multiple representations (e.g., daytime map, nighttime map, or maps created at different times). This segmentation allows the robot to select the appropriate map for current conditions, maintaining high recognition accuracy without requiring complex real-time adaptation mechanisms.

The patent applies preliminary action by creating multiple maps in advance under different illumination conditions or at different times, and storing them for later use. Instead of processing illumination changes in real-time, the system pre-processes the environment into multiple maps during different conditions, then simply selects the appropriate pre-created map during operation. This approach improves accuracy while minimizing real-time processing requirements and storage overhead.

The patent introduces maps as an intermediary between the robot's movement and location recognition. Instead of directly inferring current location from previous driving information (which fails when the robot is moved externally), the system uses pre-created maps as a reference framework. When the robot's location is forcibly changed, it can still recognize its new position by comparing current sensor data against the stored maps, which serve as a stable reference that is independent of the robot's movement history.

The patent implements periodic action by creating maps at regular intervals or under specific periodic conditions (e.g., once during daytime, once during nighttime, or at scheduled times). This periodic map creation ensures coverage of different illumination conditions without requiring continuous map updates. The system balances robustness by creating maps periodically under varying conditions while minimizing the time spent on map management by not updating maps continuously or too frequently.