Enhancing Indoor Navigation of Mobile Robots with 2D Safety Scanners through Semantic Map Optimization and Parameter Tuning

Abstract This paper presents a navigation optimization of ASTI’s ProBOT L35 TC automated guided vehicle (AGV) using only its existing 2D SICK S300 safety scanner and SLAM-based navigation stack. The research addresses low navigation efficiency in a multi-floor, factory-like laboratory, where the default configuration caused slow driving, stop–spin behavior, doorway hesitations, and frequent ‘blocked-map’ aborts. The study combines targeted map optimization—preferred lines, narrow-corridor tags, waypoints, and forbidden zones—with detailed area geometries, including doorway regions, single-robot (one-way) sections, and stop-and-wait regions in front of constrained transitions. In parallel, key navigation parameters costmap inflation radius, preferred-line width, recovery behavior, fine-positioning timeouts, and speed limitswere systematically tuned. Across more than 200 autonomous missions on two floors, the optimized configuration reduced time-weighted travel time by approximately 46.1% on the second floor, 35.88% on representative third-floor segments, and about 37.53% on multi-floor elevator-related segments. Door-related hesitations and re-planning events were decreased by ~75%, ‘blocked-map’ aborts by ~80%, and navigation failures in the tightest laboratory—Labor 313, with a 95 cm doorway and an adjacent 107 cm corridor—both narrower than the minimum widths specified in the ProBOT/proANT-L documentation—by 75%, all while remaining compliant with ISO 3691–4. These results demonstrate that careful semantic map design and systematic parameter tuning can transform conservative stop-and-go AGV behavior into smooth, reliable, an efficient navigation.