In scenarios characterized by significant curvature, such as curves, the presence of steering system hysteresis and linearization of the vehicle model may lead to understeer and steady-state errors, thereby exerting an adverse impact on the performance of the tracking control system. To address these formidable challenges, this study introduces an innovative path planning and tracking framework. This framework leverages batch informed trees to derive the reference path and employs Nonlinear Model Predictive Control (NMPC) with pre-steering to accomplish efficient tracking of the predetermined path. Additionally, a smart car experiment platform has been established for simulating and validating various scenarios, with results demonstrating improved performance in tracking accuracy and steering smoothness when compared to Traditional Model Predictive Control (TMPC). Particularly in scenarios involving substantial curvature, such as curves, the proposed framework exhibits a nearly 50% reduction in root mean square error, underscoring its enhanced performance.

In self-driving cars, object detection algorithms are becoming increasingly important, and the accurate and fast recognition of objects is critical to realize autonomous driving. The existing detection algorithms are not ideal for the detection of small objects. This paper proposes a YOLOX-based network model for multi-scale object detection tasks in complex scenes. This method adds a CBAM-G module to the backbone of the original network, which performs grouping operations on CBAM. It changes the height and width of the convolution kernel of the spatial attention module to 7 × 1 to improve the ability of the model to extract prominent features. We proposed an object-contextual feature fusion module, which can provide more semantic information and improve the perception of multi-scale objects. Finally, we considered the problem of fewer samples and less loss of small objects and introduced a scaling factor that could increase the loss of small objects to improve the detection ability of small objects. We validated the effectiveness of the proposed method on the KITTI dataset, and the mAP value was 2.46% higher than the original model. Experimental comparisons showed that our model achieved superior detection performance compared to other models.

The PointPillars algorithm can detect vehicles, pedestrians, and cyclists on the road, and is widely used in the field of environmental awareness in autonomous driving. However, its feature encoding network only uses a minimalist PointNet network for feature extraction of point cloud information, which does not consider the global context information of the point cloud, and the local structure features are not sufficiently extracted, and these feature losses can seriously affect the performance of the object detection network. To address this problem, this paper proposes an improved PointPillars algorithm named TGPP: Transformer-based Global PointPillars. After the point cloud is divided into several pillars, global context features and local structure features are extracted through a multi-head attention mechanism, so that the point cloud after feature coding has global context features and local structure features; the two-dimensional pseudo-image generated by this feature is used for feature learning using a two-dimensional convolutional neural network. Finally, the SSD detection head is used to achieve 3D object detection. It is demonstrated that the TGPP achieves an average accuracy improvement of 2.64% in the KITTI test set.

Most existing Path-Following Algorithms (PFAs) are developed for single-unit vehicles (SUVs) and rarely for articulated vehicles (AVs). Since these PFAs ignore the motion of the trailer, they may cause large tracking deviations and ride stability issues when cornering. To this end, an Adaptive Articulation Angle Preview-based Path-Following Algorithm (AAAP-PFA) is proposed for AVs. Different from previous PFAs, in this model, a simple linear vehicle dynamics model is used as the prediction model, and an offset distance calculated by an articulation angle is used as part of the preview distance. An adaptive posture control strategy is designed to trade off the trajectory tracking performance and lateral stability performance during the path-following process. Considering a large prediction mismatch caused by using a linear vehicle dynamics model, a feedback correction method is proposed to improve the robustness of the steering control. In the comparison simulation experiment with SUV-PFA, it is confirmed that the novel PFA has better adaptability to the contradictory relationship between tracking performance and lateral stability and has strong steering control robustness.

To address the issue of incorrect recognition in the automatic emergency braking (AEB) systems on curved roads, a target recognition model is proposed to obtain the road curvature and to calculate the relative lateral distance. Based on the information from the ego vehicle and the preceding vehicles, the accurate selection of the hazardous target is accomplished. After identifying the dangerous target, a control strategy based on the fusion algorithm is proposed, because the safety distance model and the Time-to-Collision (TTC) model both have their limitations and cannot ensure driving safety and comfort simultaneously. The TTC model is optimized according to the actual relative distance between two vehicles on curved roads, the graded warning strategy and braking intervention time are established by the TTC model. And then the graded braking strategy is designed according to the safety distance model. The simulation platform is built based on Carsim and Simulink for verification and analysis. The results demonstrate that the proposed AEB control strategy on curved roads can accurately and efficiently identify the target vehicles on curved roads, avoid false triggering issues, and improve the AEB system’s reliability. And effectively avoid collisions with target vehicles that are in the same lane, improving driving safety and comfort.

Analyzing the opening and closing states of eyes and mouths by detecting the driver’s face feature points is an effective method for judging driver fatigue. However, in practical engineering applications, with the expansion of user groups, the false identification problem caused by the differentiation of individual facial features of drivers is prominent, especially for people with small eyes. To solve this problem, this paper uses the Mediapipe Facemesh module to detect face feature points and designs a perception-free calibration method for setting personalized eye opening and closing threshold combined with head postures. Compared with the traditional method of setting a fixed threshold, the precision of eye state recognition is improved by 36.4%. Finally, the model deployment and post-processing compilation are completed on the Xavier vehicle chip, achieving a running speed of 34 frames per second at most, and the subjective evaluation experience of the fatigue monitoring system is significantly improved.