Augmented Reality Based Indoor Navigation System using Point Cloud Localization for Efficient Item Search in Warehouse

Locating specific items within expansive warehouse spaces poses a significant challenge, particularly when items are dispersed across different sections. Traditional 2D map-based navigation lacks depth and detail, leading to inefficiencies as workers spend considerable time navigating to find items within an expansive warehouse. The study aims to develop an AR-based navigation system using point cloud technology, using ARCore and Immersal SDK for mapping and localization. ARCore utilized to tracks the device’s position and orientation in 3D space, aiding in AR experiences by overlaying navigation cues in real-time. The Immersal SDK is used to enable spatial and visual mapping, integrating digital content with the physical environment through point cloud localization. To conduct pathfinding analysis, the navmesh in Unity is used, combined with the Catmull-Rom spline interpolation algorithm to create smooth navigation paths. To map a space, in this case, a warehouse, an immersal mapper is used, where a camera captures objects in the real world in the form of a point cloud. Each point in the point cloud reflects latitude and longitude. This point cloud will be stored in immersal cloud services for use in the localization process after the point cloud is processed through Unity software. Testing results indicate the application effectively assists navigation, offering smooth trajectory adjustments around obstacles, though it currently lacks rerouting capabilities when paths are blocked. Time efficiency testing revealed that the application significantly reduces the time taken to locate items compared to the manual method, the mean time taken for each task was consistently lower in the group that used the app compared to the group that did not. Specifically, the time reduction was more pronounced for more time-intensive tasks, such as Item 3, where the mean completion time decreased from 458.75 seconds without the app to 208.25 seconds with the app. The average distance taken for all items is consistently lower in the group using the application compared to the group not using it, the average distance for Item 3 decreased from 151.21 meters without the application to 69.3 meters with the application. The ANOVA test was conducted to show the significant effect between those using the application and those not using it. The results indicated that the $\mathbf{p}$-value was below 0.05, which significantly suggests that the application helps reduce the time and distance taken when searching for items.