FR-SLAM: A SLAM Improvement Method Based on Floor Plan Registration

In the realm of autonomous indoor navigation for mobile robots, the fusion of Simultaneous Localization and Mapping (SLAM) technologies has evolved into a vital pillar for achieving precise environmental cartography. Traditional SLAM approaches, however, frequently necessitate exhaustive exploration of indoor spaces to compile comprehensive maps, a process that prolongs path planning phases and escalates travel time to designated destinations. Furthermore, the accumulation of errors during robotic motion can compromise localization precision, thereby diminishing navigational efficiency. This research presents a novel SLAM framework, designated as FR-SLAM, which innovatively incorporates floor plan alignment via a morphology-driven algorithm for floor plan registration. This approach expedites the generation of intricate motion maps and streamlines path planning, facilitating swift traversal towards target locations. To bolster both registration accuracy and robot localization precision, a real-time updating mechanism is integrated. This mechanism dynamically synchronizes the robot’s instantaneous position with the architectural layout, dynamically refining floor plan registration in real-time to ensure pinpoint localization. Extensive evaluations, both empirical and through simulated scenarios, have highlighted the distinct advantages of this methodology compared to established algorithmic benchmarks, notably in terms of floor plan registration fidelity and reduced navigation time to target points.