Semantic Map

Semantic maps represent environments by integrating spatial information with semantic labels, aiming to provide robots and autonomous systems with a richer understanding of their surroundings than traditional occupancy grids. Current research focuses on developing robust methods for creating and updating these maps using various sensor data (LiDAR, RGB-D cameras) and incorporating large language models (LLMs) for higher-level reasoning and instruction following. This work is significant because accurate and comprehensive semantic maps are crucial for enabling advanced capabilities in robotics, autonomous navigation, and remote sensing applications, such as improved object navigation, scene understanding, and instruction-following.