This chapter introduces a real-time autonomous vehicle control framework that integrates Sliding Mode Control (SMC) with a classical Hough Transform-based lane detection pipeline. The perception module utilizes lightweight image processing operations such as grayscale transformation, Gaussian smoothing, edge extraction, and line approximation to derive lane-related features from camera input. On top of this lightweight vision module, SMC is employed to generate robust steering and velocity commands capable of handling dynamic road scenarios and visual perturbations. The system was developed and evaluated within a unity-based simulation environment and benchmarked against a conventional Proportional–Integral–Derivative (PID) controller. Experimental observations reveal that the proposed SMC approach substantially outperforms PID across multiple control dimensions. Notably, SMC achieved a lower average speed error (1.16 km/h vs. 2.93 km/h), reduced velocity fluctuation (0.14 km/h vs. 0.37 km/h), and faster disturbance recovery (0.22 s vs. 0.31 s). Furthermore, it demonstrated a significantly smoother control profile, with a Steering Smoothness Index (SSI) of 0.16, over nine times lower than PID-indicating greater stability and reduced control effort. Although classical lane detection methods such as the Hough Transform may underperform in highly unstructured environments, their low computational overhead enables fast and deterministic inference. When paired with a nonlinear robust control strategy like SMC, the overall system achieves high tracking precision (average lane deviation: 6.96 px) and operational reliability. These findings highlight the effectiveness of hybrid lightweight perception and robust control architectures for real-time autonomous driving applications, especially on embedded or resource-constrained platforms