Automotive Science and Engineering

---

In this paper, an adaptive cruise control system is designed that is controlled by a neural network model. This neural network model is trained with data resulting from the simulation of a multi-objective nonlinear predictive adaptive cruise control system. For this purpose, first, an adaptive cruise control system was designed using the concept of model predictive control based on a nonlinear model to maintain the desired speed of the driver, maintain a safe distance with the car in front, reducing fuel consumption and increasing ride comfort. Due to the time-consuming computations in predictive control systems and the consequent need for powerful and expensive hardware, it was decided to use the extracted data from the simulation of this designed cruise control system to train a neural network model and use this model to achieve control objectives instead of the predictive controller. Using the neural network model in the cruise control system, despite a significant reduction in computation time, the control objectives were well achieved, and in fact a combination of model predictive controller accuracy and neural network controller speed was used.

---

The turn-around task is one of the challenging maneuvers in automated driving which requires intricate decision making, planning and control, concomitantly. During automatic turn-around maneuver, the path curvature is too large which makes the constraints of the system severely restrain the path tracking performance. This paper highlights the path planning and control design for single and multi-point turn of autonomous vehicles. The preliminaries of the turn-around task including environment, vehicle modeling, and equipment are described. Then, a predictive approach is proposed for planning and control of the vehicle. In this approach, by taking the observation of the road and vehicle conditions into account and considering the actuator constraints in cost function, a decision is made regarding the minimum number of steering to execute turn-around. The constraints are imposed on the speed, steering angle, and their rates. Moreover, the collision avoidance with road boundaries is developed based on the GJK algorithm. According to the simulation results, the proposed system adopts the minimum number of appropriate steering commands while incorporating the constraints of the actuators and avoiding collisions. The findings demonstrate the good performance of the proposed approach in both path design and tracking for single- and multi-point turns.

---

Path-tracking for autonomous vehicles at physical handling limits is severely challenged by nonlinear tire saturation, which degrades conventional Active Front Steering (AFS) systems. This study proposes a hierarchical control architecture coordinating AFS, Torque Vectoring Control (TVC), and Active Rear Camber (ARC) to enhance path-tracking accuracy under limit driving conditions. An upper-level Linear Model Predictive Control (LMPC) algorithm is designed to calculate the virtual corrective yaw moment and the optimal rear camber angle. Simultaneously, a lower-level three-mode Quadratic Programming (QP) framework dynamically allocates torques based on instantaneous tire friction capacities. MATLAB/CarSim co-simulations of severe Double Lane Change (DLC) maneuvers validate the system's efficacy. Quantitatively, during a 140 km/h maneuver on dry asphalt, the proposed fully integrated system expands the maximum achievable lateral acceleration to 0.8g. Compared to the baseline AFS configuration, it significantly reduces the root-mean-square (RMS) and peak lateral tracking errors by 32% (to 0.239 m) and 27% (to 0.687 m), respectively, while concurrently decreasing the peak steering demand by 27%. Furthermore, under low-friction critical conditions (60 km/h, μ=0.5), the controller effectively limits sideslip oscillations and prevents vehicle spin-out. Ultimately, the formulated hierarchical framework manages the over-actuation dynamically, yielding a peak execution time that consumes only 74% of the real-time step limit, providing a highly viable and computationally efficient strategy for automotive Electronic Control Unit (ECU) implementation.