This study proposes an innovative fault diagnosis algorithm developed for the 4WIS4WID vehicle system, aiming to overcome fault location and identification challenges. The 4WIS4WID system utilizes hardware redundancy and fault-tolerant control to maintain vehicle operation even when a fault occurs. However, because the number of degrees of freedom of the vehicle is less than the number of faults, it is impossible to distinguish analytically between faults. The proposed algorithm employs residual analysis to address this limitation, capturing the differences between predicted vehicle behavior and actual sensor data. The residuals are converted into a three-channel image using a 2-D histogram and a logical function to form a single fault image. A convolutional neural network (CNN) learns these fault images to detect the occurrence of a fault and accurately determine its location. The Recursive Least Square (RLS) algorithm is utilized to classify fault types. This method identifies the fault's size and type, and the vehicle's output in which the fault occurred is estimated. The proposed algorithm's fault diagnosis and classification performance are verified through CarMaker-based simulation. This systematic approach to fault diagnosis enhances vehicle safety and reliability in intelligent vehicle applications.

This paper proposes a novel lane change path optimization algorithm for emergency collision avoidance in intelligent vehicles. Traditional path planning often faces challenges such as optimizing many variables, being unsuitable for emergency scenarios, or producing paths that are difficult for vehicles to follow accurately. Our approach employs lagged longitudinal and lagged curvature models with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$G^2$</tex-math></inline-formula> continuous curves, generating reliable collision avoidance trajectories under low-level actuator time lag. The lagged longitudinal model is designed to be differentiable in the spatial domain, allowing curvature optimization in the curvilinear coordinates. The lagged curvature model facilitates rapid lane changes, making it suitable for emergency collision avoidance. The proposed method optimizes only dual connecting points and enhances computational efficiency using analytic gradient-based nonlinear programming. Comparisons with existing path planning methods–namely, the fifth-order spline optimization method, model predictive control-based optimization, and the time-domain fifth-order polynomial optimization method–on straight and curved roads demonstrate that the proposed method minimizes collision avoidance areas more effectively. The proposed path is verified with a simple tracking controller in high-fidelity simulation using TruckSim.

This study proposes an innovative fault diagnosis algorithm developed for the 4WIS4WID vehicle system, aiming to overcome fault location and identification challenges. The 4WIS4WID system utilizes hardware redundancy and fault-tolerant control to maintain vehicle operation even when a fault occurs. However, because the number of degrees of freedom of the vehicle is less than the number of faults, it is impossible to distinguish analytically between faults. The proposed algorithm employs residual analysis to address this limitation, capturing the differences between predicted vehicle behavior and actual sensor data. The residuals are converted into a three-channel image using a 2-D histogram and a logical function to form a single fault image. A convolutional neural network (CNN) learns these fault images to detect the occurrence of a fault and accurately determine its location. The Recursive Least Square (RLS) algorithm is utilized to classify fault types. This method identifies the fault's size and type, and the vehicle's output in which the fault occurred is estimated. The proposed algorithm's fault diagnosis and classification performance are verified through CarMaker-based simulation. This systematic approach to fault diagnosis enhances vehicle safety and reliability in intelligent vehicle applications.

This paper proposes a novel lane change path optimization algorithm for emergency collision avoidance in intelligent vehicles. Traditional path planning often faces challenges such as optimizing many variables, being unsuitable for emergency scenarios, or producing paths that are difficult for vehicles to follow accurately. Our approach employs lagged longitudinal and lagged curvature models with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$G^2$</tex-math></inline-formula> continuous curves, generating reliable collision avoidance trajectories under low-level actuator time lag. The lagged longitudinal model is designed to be differentiable in the spatial domain, allowing curvature optimization in the curvilinear coordinates. The lagged curvature model facilitates rapid lane changes, making it suitable for emergency collision avoidance. The proposed method optimizes only dual connecting points and enhances computational efficiency using analytic gradient-based nonlinear programming. Comparisons with existing path planning methods–namely, the fifth-order spline optimization method, model predictive control-based optimization, and the time-domain fifth-order polynomial optimization method–on straight and curved roads demonstrate that the proposed method minimizes collision avoidance areas more effectively. The proposed path is verified with a simple tracking controller in high-fidelity simulation using TruckSim.

Highly autonomous driving technology is expected to improve driving safety and convenience, and collision avoidance technology is essential for fully autonomous driving. Planning a collision-free trajectory that includes velocity and path is one of the most challenging objectives. Optimization-based trajectory planners have been proposed in many previous studies because they offer a high degree of freedom and can handle various situations. However, most previous trajectory planners used nonlinear programming due to the nonlinearity or non-convexity of the optimization problem. These methods come with a high computational load. The trajectory planner requires the real-time ability to cope with dynamically changing environments. This article focuses on the trajectory planning of autonomous vehicles through quadratic programming (QP), which requires a low computational load. To achieve this, we introduce the longitudinal-lateral decomposition method. In addition, collision-free constraints are expressed as linear constraints through proposed ingenious dual functions. The proposed weak duality optimization problem has a QP form and optimized trajectory and obstacle avoidance timing through only one QP problem. This study verified that the proposed trajectory planner could plan smooth collision-free maneuvers for several driving situations by simulations.

Abstract To address the computational challenges of Model Predictive Control (MPC), recent research has studied using imitation learning to approximate MPC with a computationally efficient Deep Neural Network (DNN). However, this introduces a common issue in learning-based control, the simulation-to-reality (sim-to-real) gap. Inspired by Robust Tube MPC, this study proposes a new control framework that addresses this issue from a control perspective. The framework ensures the DNN operates in the same environment as the source domain, addressing the sim-to-real gap with great data collection efficiency. Moreover, an input refinement governor is introduced to address the DNN’s inability to adapt to variations in model parameters, enabling the system to satisfy MPC constraints more robustly under parameter-changing conditions. The proposed framework was validated through two case studies: cart-pole control and vehicle collision avoidance control, which analyzed the principles of the proposed framework in detail and demonstrated its application to a vehicle control case.

The rapid advancement of autonomous driving and ADAS technologies has increased the demand for high-definition (HD) lane-level maps that accurately preserve rich geometric information while scaling to city-wide coverage. While traditional polyline-based formats are widely used, they struggle to provide continuous representations of key geometric properties such as curvature and heading angle, which are essential for autonomous driving applications. Curve-based representations have been introduced to address these limitations, but existing methods are often restricted to simplified or sparsely connected road networks, limiting their effectiveness in large-scale, real-world environments. This study presents an arc-spline-based lane representation framework that efficiently models complex, large-sized lane-level maps while preserving continuous road geometry. To achieve this, we introduce a novel road network decomposition and merging method that enables structured parameterization without requiring full map-scale optimization. Instead, optimization is localized to cluster connection regions, significantly enhancing computational efficiency. Validation using lanelet maps from the nuScenes dataset demonstrates that our approach maintains an average approximation error of 4.3 cm while preserving detailed lane topology and global tangential continuity, while also achieving a significant reduction in storage requirements compared to conventional polyline formats.