Multi-LiDAR SLAM and intelligent navigation for autonomous logistic robots in ROS2 environments

Abstract

Traditional automated guided vehicles (AGVs) are restricted by their reliance on predefined paths, limiting adaptability in dynamic warehouse environments. While autonomous mobile robots (AMRs) overcome this limitation through on-board simultaneous localization and mapping (SLAM) and autonomous navigation, standard configurations often suffer from top mounted-sensor blind spots when loads are carried on the chassis. To address these coverage gaps, an indoor logistic AMR based on the robot operating system 2 (ROS2) was designed and evaluated. The platform was developed by combining a multi-LiDAR perception stack with low-cost industrial actuation and a lightweight, fleet-style user interface. Within the system architecture, data from two light detection and ranging (LiDAR) sensors were merged at the topic level into a single virtual scan for SLAM toolbox and Nav2. Additionally, actuation and wheel odometry were driven by an RS-485 Modbus-based brushless DC (BLDC) motor controller, while ultrasonic sensors for short-range safety, an inertial measurement unit (IMU) for orientation, and a network of microcontroller calling stations communicating via message queuing telemetry transport (MQTT) were integrated into the platform. Experimental validation demonstrated successful multi-LiDAR fusion, with the Modbus motor driver achieving a motion-control error of 0.36 % and a speed-retrieval error of 0.43 %. Furthermore, calling-station commands were reliably executed over MQTT, and a point-to-point navigational repeatability of 10.3 cm was achieved. These findings indicate that an integrated multi-LiDAR ROS2 AMR provides a highly practical solution for indoor logistics. Through the proposed sensor merger and calling-station handshake, two recurring vulnerabilities of standard ROS2 deployments—single-LiDAR coverage gaps and Nav2 goal-overwriting behavior—were successfully resolved.