The market for autonomous vehicles is exploding. Self-driving cars are now capable of navigating complex urban environments without the need for artificial landmarks. In contrast, the industry standard for AGV localization still relies upon meticulously surveyed reflectors to estimate vehicle pose. This is in spite of having the advantage of operating indoors, and in far less dynamic environments. While effective, reflector-based solutions are both costly and inflexible. We can do better.
AGV3D is LPR's drop-in localization solution for AGVs. It employs the latest in 3D LIDAR technology to segment natural features as landmarks. This eliminates the need to build out additional infrastructure such as guidelines, or reflectors. The end result is an AGV localization system that provides increased flexibility at lower cost.
Data gathered from a supervised walk-through of the warehouse is used to create a high-fidelity, 3D reconstruction of the space.
Landmark Map Generation
Offline feature segmentation enables the creation of a 3D landmark map using natural features from the warehouse environment.
Online Map-based Localization
3D features segmented at runtime are correlated to the landmark map. Detected landmarks are used to compute an estimate of the AGV's pose.