As a link connecting the environmental perception system and the decision-making system, accurate obstacle trajectory prediction provides a reliable guarantee of correct decision-making by autonomous vehicles. Oriented toward a mixed human-driven and machine-driven traffic environment, a vehicle trajectory prediction algorithm based on an encoding–decoding framework composed of a multiple-attention mechanism is proposed. Firstly, a directed graph is used to describe vehicle–vehicle motion dependencies. Then, by calculating the repulsive force between vehicles using a priori edge information based on the artificial potential field theory, vehicle–vehicle interaction coefficients are extracted via a graph attention mechanism (GAT). Subsequently, after concatenating the vehicle–vehicle interaction feature with the encoded vehicle trajectory vectors, a spatio-temporal attention mechanism is applied to determine the coupling relationship of hidden vectors. Finally, the predicted trajectory is generated by a gated recurrent unit (GRU) decoder. The training and evaluation of the proposed model were conducted on the NGSIM public dataset. The test results demonstrated that compared with existing baseline models, our approach has fewer prediction errors and better robustness. In addition, introducing artificial potential fields into the attention mechanism causes the model to have better interpretability.

Humanlike driving is significant in improving the safety and comfort of automated vehicles. This paper proposes a personalized motion planning method with driver characteristics in longitudinal and lateral directions for highway automated driving. The motion planning is decoupled into path optimization and speed optimization under the framework of the Baidu Apollo EM motion planner. For modeling driver behavior in the longitudinal direction, a car-following model is developed and integrated into the speed optimizer based on a weight ratio hypothesis model of the objective functional, whose parameters are obtained by Bayesian optimization and leave-one-out cross validation using the driving data. For modeling driver behavior in the lateral direction, a Bayesian network (BN), which maps the physical states of the ego vehicle and surrounding vehicles and the lateral intentions of the surrounding vehicles to the driver’s lateral intentions, is built in an efficient and lightweight way using driving data. Further, a personalized reference trajectory decider is developed based on the BN, considering traffic regulations, the driver’s preference, and the costs of the trajectories. According to the actual traffic scenarios in the driving data, a simulation is constructed, and the results validate the human likeness of the proposed motion planning method.

Scenario-based validation requires efficient screening of large test libraries, but existing complexity metrics often combine traffic hazard and automated-driving-system workload into a single construct. This study proposes a two-dimensional framework that separates scenario hazardousness from system processing difficulty. Hazardousness combines normalized acceleration and time-to-collision measures. Processing difficulty aggregates perception, decision-making, and execution demands, with weights obtained from a triangular fuzzy analytic hierarchy process. The framework was compared with four baseline metrics using 2000 Latin-hypercube-sampled cut-in scenarios and was explored in 22 closed-field scenarios tested with three production Level 2 vehicles. The proposed metric produced a broad and less concentrated distribution of cases across the normalized score range. However, total complexity was not significantly correlated with any vehicle score in the 22-case dataset; therefore, the field experiment supports only preliminary discriminant analysis, not predictive validity. A 10% hierarchical-weight perturbation analysis showed that the three largest global weights remained the same set in 82.3% of 10,000 simulations. The model is currently limited by empirically assigned perception constants, a small field sample, few ultra-high-complexity cases, and omission of road curvature, crosswinds, tire condition, and actuator delay. The additive aggregation assumes that interactions among criteria can be neglected for the intended comparison; this assumption has not been validated. The framework is best viewed as an interpretable scenario-screening tool requiring further calibration and validation.

Centralized platoon control strategies typically face substantial computational demands and exhibit limited disturbance-rejection capability. To address these challenges, this study adopts a hierarchical leader–follower architecture and develops a hybrid longitudinal control framework that integrates a supervisory layer with a calibration-based execution layer. The leader vehicle employs model predictive control (MPC) to generate optimal reference acceleration, whereas the follower vehicles utilize fuzzy PID control to achieve stable and scalable coordination. At the execution layer, a three-dimensional throttle–brake calibration map is established to convert the desired acceleration into precise pedal commands, thereby ensuring consistent actuation characteristics across the fleet. A model-in-the-loop (MIL) simulation platform is constructed using a CarSim–VeriStand co-simulation environment, enabling high-fidelity vehicle-dynamics modeling and real-time interaction with the control system. Two representative scenarios—constant-speed cruising and following a human-driven vehicle (HDV) with significant speed fluctuations, both initialized with large spacing offsets—are evaluated. Simulation results demonstrate that, even under aggressive HDV disturbances, inter-vehicle spacing errors remain exceptionally small (maximum ≤ 0.08 m) and rapidly converge to the desired headway after startup or perturbations. Under cruising conditions, the platoon promptly reaches the target speed and eliminates initial spacing inconsistencies. Overall, the proposed hybrid strategy exhibits strong robustness and precise spacing control across diverse driving scenarios.