Indoor Positioning Systems (IPS) are important for many uses, like helping people find their way, tracking items, and providing services based on position. Traditional IPS methods, like trilateration and fingerprints, often don’t work well because of external issues such as signal disturbance, multipath effects, and signal weakening. To solve these challenges, this study studies the merging of machine learning (ML) methods with Received Signal Strength (RSS)-based tracking techniques to optimize the accuracy and reliability of IPS. The study aims to use machine learning (ML) models to improve the accuracy of RSS-based techniques like fingerprints, trilateration, and time-of-arrival. The goal is to lower mistakes and make these methods better suited for changing indoor spaces. Multiple machine learning methods are used to understand how RSS data relates to the actual positions of devices indoors. These methods include supervised learning like Support Vector Machines, k-Nearest Neighbors, and Decision Trees, unsupervised learning like K-Means clustering, and reinforcement learning like Q-learning. The results show that the mix method greatly increases placing accuracy compared to regular methods. Notably, the inclusion of ML methods allows flexible learning, allowing the system to optimize itself over time as new data is gathered.