LEE- Otomotiv-Yüksek Lisans
Bu koleksiyon için kalıcı URI
Gözat
Konu "automotive engineering" ile LEE- Otomotiv-Yüksek Lisans'a göz atma
Sayfa başına sonuç
Sıralama Seçenekleri
-
ÖgeModel predictive control and differential braking based steering redundancy for an autonomous vehicle(ITU Graduate School, 2025) Zaman, Başar ; Akalın, Özgen ; 503211718 ; AutomotiveWith the widespread adoption of autonomous driving technologies, traditional mechanical systems in vehicles are increasingly being replaced by electronic and software-based systems. One of the most prominent examples of this transformation is observed in steering systems. Steer-by-wire systems, which replace conventional mechanical steering linkages, transmit steering commands via electronic signals without any physical connection, thereby enabling the steering task. This innovative structure provides significant advantages such as design flexibility, weight reduction, more precise control capabilities, and high integration with advanced driver assistance systems. However, in the event of an electrical or software failure in the steering system, the risk of a complete loss of steering capability arises. Such a condition may not only compromise the safety of autonomous vehicles but also pose serious threats to passenger safety. Therefore, steering redundancy strategies involving alternative steering methods for steer-by-wire systems are of great importance. This thesis proposes a novel differential braking-based steering redundancy method supported by Model Predictive Control (MPC), which aims to maintain vehicle controllability under potential failure scenarios in steer-by-wire systems. In this method, steering moments are generated by applying different magnitudes of braking forces to the left and right wheels, thereby creating an alternative steering input. Through this approach, the vehicle can maintain its ability to follow the desired path even when the steer-by-wire system is nonfunctional. The proposed control architecture is designed as a two-layer hierarchical structure. In the upper layer, two PID controllers operate based on the lateral position error of the vehicle's center of gravity and the yaw angle error. These controllers generate reference values for the steering angle and curvature required for safe driving. These reference values are transmitted to the lower control layer, where an MPC based controller computes the optimal braking forces for the left and right wheels. The MPC controller predicts the future behavior of the vehicle model over a defined time horizon and generates control decisions that ensure both vehicle stability and path tracking. The calculated braking forces are then sent to the ABS controller, which applies them in a way that prevents wheel lock while generating the desired yaw moment. In the literature, differential braking is commonly used to enhance vehicle stability or correct deviations in sideslip angle or yaw rate. However, in this study, differential braking has been redefined as a primary steering redundancy mechanism, and physical effects such as self-aligning moment and scrub radius arising from tire-road interactions are also incorporated into the optimization. In this way, the vehicle can be actively steered solely through braking forces, even in the absence of a functioning steering system. This integrated approach enhances system reliability and increases the fault tolerance of steer-by-wire architectures. To evaluate the effectiveness of the proposed method, simulations were conducted using the MSC ADAMS/Simulink co-simulation environment. The FED-Alpha, a high-performance military prototype vehicle model, was used for this purpose. Simulation scenarios included NATO Double Lane Change (DLC) maneuvers performed at both low speed (35 km/h) and high speed (60 km/h). In these scenarios, it was assumed that the steer-by-wire system was deactivated, and the vehicle was required to follow the reference trajectory using only differential braking control. Two different steering models were compared within the simulation environment. The first is a simplified model that represents the steering system as a basic dynamic structure using only general stiffness, damping, and mass parameters. The second is a detailed model that includes physical effects such as the mass and inertia of subcomponents, the position of the center of gravity, scrub radius, and self-aligning moment to more accurately represent the steering dynamics. The results showed that under low-speed conditions, both models provided similar path-following performance. However, under high-speed maneuvers, the detailed model exhibited a clear advantage in terms of both trajectory tracking and maintaining vehicle stability. This model enabled more accurate and timely braking force estimations, prevented loss of balance, and ensured more effective coordination with the ABS system. These findings confirm that differential braking-based steering redundancy can be successfully applied even at high speeds. In conclusion, this thesis demonstrates that differential braking can be used as a primary steering mechanism to enhance the safety of steer-by-wire systems. Through the MPC algorithm, steering can be achieved using only braking forces, allowing the vehicle to maintain operational capability even in failure scenarios. Furthermore, the integration of detailed physical parameters such as self-aligning moment and scrub radius into the control system significantly improves control performance. This study is particularly important for the development of fail-safe vehicle systems, as it highlights how advanced control strategies can be integrated into automotive applications. Future studies may focus on real-time embedded implementation, adaptive control techniques, and hardware-in-the-loop testing to further validate the approach. At the same time, the integration of a more realistic steering mechanism into the model is also planned. In this context, not only the general stiffness and damping coefficients but also the detailed dynamic characteristics of all steering system components will be considered. In particular, the rotational inertia of the steering assembly, wheel–tire system inertia, and the inertial properties of connecting elements such as tie rods and bushings will be separately calculated and included in the model. Through this more comprehensive modeling, the system's response under real-world failure conditions will be more accurately simulated, and controller behavior will be improved accordingly. Additionally, nonlinear effects such as tire relaxation length will be incorporated, as they play a significant role in tire force generation, especially during high-speed maneuvers. With the inclusion of such factors, the controller's performance particularly under demanding dynamic scenarios will be significantly enhanced. In addition, nonlinear effects arising from tire–road interaction, such as tire relaxation length, are also planned to be incorporated into the modeling. These effects play a critical role in the generation of tire forces, especially during high-speed maneuvers, and therefore have a direct impact on the performance of the controller. In this context, by achieving a more accurate representation of tire dynamics and enhancing the physical realism of the steering system, the accuracy and stability of the control system under high-speed scenarios are expected to improve significantly.