Toward efficient and agent-scalable indoor pathfinding: Intelligent navigation-ability-driven indoor map generation using building information model

Pathfinding is key in fields like robotics, automation, and computer-aided design. Indoor maps, crucial for pathfinding, are typically generated using Building Information Modeling (BIM). However, current methods are often tailored to specific agents, like pedestrians or vehicles, limiting their adaptability. This approach requires new map generation methods each time a new type of navigation agent, such as indoor robots, is introduced. As more smart building robots emerge, there's a growing need for scalable indoor map generation schemes. This study proposes an agent-scalable indoor map generation scheme and an efficient pathfinding algorithm using BIM data. First, an indoor navigation ability model is created to represent the occupation spaces and movement capabilities of navigation agents. Second, a navigation-ability-driven indoor map generation scheme is designed, using BIM data to create voxel-based passable areas and continuous span areas, ensuring scalability and universality. This results in a simple polygon outline and indoor triangular map. Third, an efficient pathfinding scheme is developed to navigate these generated indoor maps. Finally, experiments are conducted on various BIM models and agents. Experimental results show that the proposed scheme is adaptable to various agents and scales for those with different navigation abilities. The pathfinding algorithm finds the shortest path in an average of 0.4 s in a one-floor shopping plaza covering over 40,000 m². The scalability of this indoor map generation scheme allows new indoor navigation agents to access customized maps based on their navigation parameters, paving the way for more intelligent applications in smart buildings.