Lane change maneuvers are critical for safety and robustness of automated driving systems. Yet, simulation environments like CARLA still rely on basic path planners that often yield to unrealistic and dynamically inconsistent trajectories, particularly on curved roads. In this work, we propose an adaptive B-spline-based trajectory planning framework for lane change execution in CARLA. Our method generates smooth, curvature-aware trajectories by leveraging road waypoints and real-world derived maneuver parameters such as speed, direction, and duration. A tuned PID controller handles both longitudinal and lateral motion, ensuring accurate and stable trajectory tracking. We evaluated our approach through 400 simulations with various road geometries and directions. Compared to CARLA’s default planner, our approach achieves significant improvements: lateral jerk is reduced by 76.4%, longitudinal jerk by 89.8%, heading error by 39.2%, and lateral displacement error by 93.5%, bringing our model within the limits of the state of the art control benchmarks and passenger-comfort guidelines.