Design and Performance Evaluation of an Autonomous Mobile Robot for Warehouse Automation

Abstract

Warehouse automation increasingly requires mobile systems that can operate safely in changing layouts rather than follow fixed guide paths. This paper presents the design and simulation-based performance evaluation of a differential-drive autonomous mobile robot (AMR) for pallet-light and tote transport in indoor warehouses. The proposed architecture integrates two-dimensional LiDAR, wheel encoders and an inertial measurement unit with occupancy-grid mapping, Adaptive Monte Carlo Localization, A* global path planning, and a Dynamic Window Approach local planner. A safety supervisor limits velocity near obstacles and triggers emergency stops when clearance falls below a defined threshold. The methodology evaluates the robot in three warehouse conditions: static aisles, mixed human/robot traffic, and dense dynamic traffic. Sixty seeded missions were executed in a reproducible 2-D test environment and were assessed using path length, mission time, localization error, success rate, and collision rate. The proposed AMR completed 100%, 95%, and 90% of missions in the three scenarios, respectively, while maintaining mean localization error below 0.10 m. Compared with a scripted fixed-route AGV baseline, the AMR reduced mean mission time by approximately 26-31% under dynamic conditions. The study demonstrates that combining probabilistic localization with layered global-local planning provides a practical methodology for flexible warehouse automation, while also identifying the need for hardware validation, fleet-level coordination, and standardized safety testing before industrial deployment.