Comparative Analysis of Machine Learning-Based Indoor Positioning for AMRs Using Wi-Fi RSSI Data

The development of low-cost and high-precision positioning systems in industrial environments is crucial for the efficient operation of autonomous robots. Wi-Fi-based positioning offers a viable alternative by leveraging existing infrastructure and minimizing the need for additional hardware. In this study, we applied and compared various machine learning methods for indoor positioning of autonomous mobile robots based on Wi-Fi signals. The experimental results indicate that Decision Tree and Random Forest algorithms achieve the lowest positioning errors, whereas regression-based methods are proved to be insufficient. This study contributes to the literature by providing comparative analysis of various machine learning approaches, validated through real-world experiments. Future research will focus on real-time updating the database and integrating additional sensors into the positioning system to mitigate the impact of environmental variability on signal fluctuations.