26 March 2026 in IEEE Transactions on Geoscience and Remote Sensing
Lane detection is a fundamental task in computer vision with broad applications in autonomous driving. Recent advances have focused on line-anchor-based representations for their efficiency and high accuracy. However, most existing methods adopt anchors that are continuous in point sampling yet discrete in width and fixed in direction. This leads to geometric misalignments with real-world lane lines, which exhibit continuous width representation and smoothly varying directions. These misalignments cause Symmetric Point Ambiguity and Magnified Localization Errors, ultimately degrading both evaluation reliability and detection accuracy. To address these issues, we propose Angle-Aware Rectangle Anchors (ARA), a novel representation with continuous width and adaptive directional alignment, that effectively captures the geometric variation of lane lines. In addition, we propose the Three-phase Angle-Thresholded Line-Area Transition (TALAT) Loss, which dynamically switches between loss formulations based on angular thresholds and overlap quality, enabling a smooth transition from coarse to fine supervision. Extensive experiments on multiple standard benchmarks (TuSimple, CULane, CurveLanes, and LLAMAS) demonstrate that our approach achieves competitive or state-of-the-art (SOTA) performance. Notably, ARA exhibits superior robustness in challenging scenarios, while maintaining real-time inference speed. Our code and models are publicly available at https://github.com/changehome717/ARA-Lane-Detection.
16 October 2025 in IEEE Internet of Things Journal
Accurate long-term vehicle trajectory prediction is crucial for building intelligent transportation systems and facilitating decision-making in autonomous driving. Existing trajectory prediction methods assume that all vehicles are intelligent connected vehicles (ICVs) and utilize historical trajectories to conduct predictions. However, they neglect the mixed traffic conditions involving human-driven vehicles (HDVs), whose operation data are unavailable, and often persist for the foreseeable future in real-world scenarios. When ICVs need to predict the trajectories of the HDVs around, they cannot use HDVs’ historical position information due to the lack of intelligent connected devices and can only use their own trajectories for indirect prediction. To address this, a temporal causal convolutional bidirectional recurrent network (TCCBRN) is proposed, which combines causal convolution layers and bidirectional recurrent neural networks (RNNs). The graph-based causal convolution layer is used for processing trajectory data and extracting spatial interaction features, and the recurrent neural layer achieves indirect predictions based on these features. Thus, the proposed network infers the future motion of surrounding HDVs by leveraging only the trajectories of nearby ICVs. Additionally, we introduce an improved dung beetle optimization (IDBO) algorithm to optimize the high-dimensional hyperparameters of the TCCBRN. By introducing piecewise chaotic mapping and dynamic weight coefficients to enhance the population generation and updating mechanisms, IDBO increases both the optimization dimension and search efficiency, making it well-suited for hyperparameter optimization in multiagent surrounding prediction tasks. Cross-scene experiments on four datasets show that our method outperforms state-of-the-art baselines for predicting HDV trajectories from surrounding ICV data, achieving accuracy levels that support practical application.