This work focuses on developing a deep learning-based robotic navigation system capable of indoor localization and autonomous movement. It integrates various machine learning approaches, including Convolutional Neural Networks (CNNs), Deep Neural Networks (DNNs), and Region-based CNNs (RCNNs), each utilized for tasks such as feature extraction, object detection, and obstacle avoidance. In addition to these, conventional techniques like Simultaneous Localization and Mapping (SLAM), Random Forests, and Support Vector Machines (SVMs) are incorporated to enhance environmental mapping and facilitate efficient decision-making. The system is trained using Kaggle’s Internal Navigation Dataset for Systems, which leverages wireless signal data to estimate and refine the robot’s indoor position. The entire framework operates within a simulated environment, assuming robust and scalable mechanisms for localization and path planning to address the inherent limitations of wireless-based location tracking. By integrating deep learning with traditional methods, the system effectively achieves real-time indoor navigation in a sustainable manner. Performance metrics revealed that the CNN model produced a Mean Squared Error (MSE) of 1023.39, the DNN model achieved an MSE of 0.32502, and the RCNN model recorded a Root Mean Square Error (RMSE) of 87.93. Among traditional methods, SVMs achieved an accuracy of 88.46%, while Random Forests performed the best with an accuracy of 96.15%. SLAM recorded an RMSE of 7.4798, highlighting its effectiveness in spatial mapping.