The rapid advancement of autonomous cars, however, demands reliable sensor operations to ensure safety, efficiency, and reliability. This research focuses on designing a real-time sensor fault detection framework for vital sensors like accelerator, brake pressure, and steering sensors. From an analysis of time series data generated from synthetic data offered by the CARLA simulator, as well as real-world sources such as the A2D2 dataset, this research explores state-of-the-art deep learning and machine learning methods, such as transformer-based anomaly detection, 1D-CNN, SVM, and LSTM-autoencoders. The objective is to detect and correct sensor anomalies to allow for timely interventions that can maximize vehicle operating reliability and safety. The suggested framework is designed to address the risks caused by faulty sensor data, which makes the technology more appealing to the masses. Finally, this research goes a long way in improving the resilience of autonomous vehicle systems, and hence towards more reliable and safe transportation systems.