Development of an Autonomous Mobile Robot System Capable of Tracking Workers in a Factory Environment

Abstract

An autonomous mobile robot (AMR) that follows a worker can reduce the physical burden on the worker, but stable navigation remains difficult in the narrow and cluttered spaces of a factory, where the conventional dynamic window approach (DWA) tends to prefer high linear velocities and therefore turns with a large radius. This paper proposes a worker-following navigation system with three specific elements: a unique ArUco marker whose pose, obtained by solving the perspective-n-point problem, is transformed into the map frame to define the goal point; an A* search whose evaluation function includes the costmap cost, followed by waypoint pruning and cubic B-spline refinement of the resulting path; and an improved DWA that adds a corner-aware term regulating the linear velocity according to the heading change expected over the prediction horizon. The system is implemented on an AMR equipped with a 3-D light detection and ranging (LiDAR) and a camera, and is evaluated in a demonstration factory through two scenarios of 30 trials each. The proposed algorithm completes all 60 trials without a collision, whereas Nav2 and the fast marching method (FMM)-DWA record 22 and 17 collisions, respectively. The experimental results demonstrate that path-geometry-aware velocity control improves navigation stability and reduces collision risk in complex factory environments.