Multi-Sensor Localization for End-to-End Logistic Robots Across Indoor and Outdoor Areas

Logistics robots are becoming increasingly necessary for practical scenarios such as logistics terminal distribution and goods transportation within campuses or factories. These robots must be able to provide end-to-end delivery services in indoor and outdoor settings, including cross-floor environments. However, single-sensor localization methods are insufficient for such complex environments. To address this, we propose a multi-sensor localization method that fuses encoders, IMU (Inertial Measurement Unit), lidar, camera, and GPS measurements using EKF(Extended Kalman Filter) and AMCL(Adaptive Monte Carlo Localization) algorithm. Additionally, we introduce a map structure and management method to support end-to-end robots operating in a campus. Our approach has successfully solved the localization challenges of logistics robots in the campus environments discussed in this paper, demonstrating advantages over existing localization methods.