Standard AUV simulations often assume idealized, disturbance-free conditions, making it difficult to validate controllers for real-world deployments. For my RBE 502 Robotic Controls final project, my partner and I developed a physically grounded, stochastic ocean current simulator for a REMUS 100 AUV using the MSS MATLAB toolbox to test controllers against dynamic, Ekman-spiral-modeled wind and depth conditions. Realizing that standard heading controllers fail to account for cross-currents—resulting in over a kilometer of lateral drift during a standard run, we modified the existing MSS controllers by building a sideslip-compensating course autopilot. This system uses onboard IMU measurements to estimate drift and actively crab the vehicle into the current, ensuring it maintains a true ground-track course rather than just a fixed body angle. By prioritizing environmental compensation and strategic course control, we reduced simulated lateral tracking errors by over 99% in stormy conditions, proving the critical necessity of designing marine control systems for non-ideal, real-world hydrodynamics.