The premise of vehicle intelligent decision making is to obtain vehicle motion state parameters accurately and in real-time. Several state parameters cannot be measured directly by vehicle sensors, so estimation algorithms based on filtering are effective solutions. The most representative algorithm is the Kalman filter, especially the standard unscented Kalman filter (UKF) that has been widely used in vehicle state estimation because of its superiority in dealing with nonlinear filtering problems. However, although the UKF assumes that the noise statistics of the system are known, due to the complex and changeable operating conditions, sensor aging and other factors, these noises vary. In order to realize high-precision vehicle state estimation, a noise-adaptive UKF algorithm is proposed in this article. The maximum a posteriori (MAP) algorithm is used to dynamically update the noise of the vehicle system, and it is embedded into the update step of the UKF to form an adaptive unscented Kalman filter (AUKF). The system will dynamically update the noise when noise statistics are unknown and prevent filter divergence by adjusting the mean and covariance of the estimated noise to improve accuracy. On this basis, the proposed method is verified by the joint simulation of CarSim and Matlab/Simulink, confirming that the AUKF performs better than the standard UKF in estimation accuracy and stability under different degrees of noise disturbance, and the estimation accuracy for the yaw rate, side slip angle and longitudinal velocity is improved by 20.08%, 40.98% and 89.91%, respectively.

The curvature of the lane output by the vision sensor caused by shadows, changes in lighting and line breaking jumps over in a period of time, which leads to serious problems for unmanned driving control. It is particularly important to predict or compensate the real lane in real-time during sensor jumps. This paper presents a lane compensation method based on multi-sensor fusion of global positioning system (GPS), inertial measurement unit (IMU) and vision sensors. In order to compensate the lane, the cubic polynomial function of the longitudinal distance is selected as the lane model. In this method, a Kalman filter is used to estimate vehicle velocity and yaw angle by GPS and IMU measurements, and a vehicle kinematics model is established to describe vehicle motion. It uses the geometric relationship between vehicle and relative lane motion at the current moment to solve the coefficient of the lane polynomial at the next moment. The simulation and vehicle test results show that the prediction information can compensate for the failure of the vision sensor, and has good real-time, robustness and accuracy.

This paper investigates the multi-objective optimization of the peak–valley difference, operating cost, and carbon emissions for urban logistics battery-swapping stations (BSSs) under photovoltaic uncertainty and stochastic demand. Unlike conventional plug-in charging, battery swapping decouples energy replenishment from the vehicle dwell time, enabling rapid service, but introducing discrete swap arrivals and power–inventory coupling challenges that continuous-load models cannot capture. A Stackelberg game-based framework models grid–BSS interactions, where the grid acts as the leader by setting time-of-use prices and BSSs respond by optimizing charging/discharging schedules. Carbon emissions are quantified using real-time carbon intensity data obtained from the Electricity Maps platform. The battery-swapping demand is modeled as a Poisson process, and a unified power–inventory coupling model captures the bidirectional dependence among PV generation, grid purchases, energy storage operations, and battery inventory dynamics, where the inventory feasibility constrains the power decisions. For multi-station coordination, an adaptive ADMM decomposes the problem into parallelizable sub-problems. Case studies of a 49-vehicle fleet across three BSSs in Qingdao, China, show that, compared with a no-optimization baseline, the proposed method reduces the peak–valley difference by approximately 21.6%, the operating cost by approximately 10.2%, and carbon emissions by approximately 15.7%. Compared with the single-objective counterparts, the multi-objective formulation further improves the peak–valley difference by approximately 26.9% and increases emission reduction by approximately 16.9%; paired t-tests on repeated runs indicate statistical significance (p < 0.05). The framework provides a scalable methodology for low-carbon BSS scheduling with explicit power–inventory coupling.