The multi-strategy optimized dream optimization algorithm (MSDOA) is proposed to address the challenges of inadequate search capability, slow convergence, and susceptibility to local optima in intelligent optimization algorithms applied to UAV three-dimensional path planning, aiming to enhance the global search efficiency and accuracy of UAV path planning algorithms in 3D environments. First, the algorithm utilizes Bernoulli chaotic mapping for population initialization to widen individual search ranges and enhance population diversity. Subsequently, an adaptive perturbation mechanism is incorporated during the exploration phase along with a lens imaging reverse learning strategy to update the population, thereby improving the exploration ability and accelerating convergence while mitigating premature convergence. Lastly, an Adaptive Individual-level Mixed Strategy (AIMS) is developed to conduct a more flexible search process and enhance the algorithm’s global search capability. The performance of the algorithm is evaluated through simulation experiments using the CEC2017 benchmark test functions. The results indicate that the proposed algorithm achieves superior optimization accuracy, faster convergence speed, and enhanced robustness compared to other swarm intelligence algorithms. Specifically, MSDOA ranks first on 28 out of 29 benchmark functions in the CEC2017 test suite, demonstrating its outstanding global search capability and conver-gence performance. Furthermore, UAV path planning simulation experiments conducted across multiple scenario models show that MSDOA exhibits stronger adaptability to complex three-dimensional environments. In the most challenging scenario, compared to the standard DOA, MSDOA reduces the best cost function fitness by 9% and decreases the average cost function fitness by 12%, thereby generating more efficient, smoother, and higher-quality flight paths.
12 August 2024 in Service Oriented Computing and Applications
Monocular 3D vehicle shape and pose estimation is of great significance for scene understanding in autonomous driv-ing. However, it encounters challenges owing to the lack of depth information, complex interactions between objects, and efficient modeling requirements. Unlike existing monocular methods that often rely on either coarse PCA-based shape priors or computationally expensive attention mechanisms, this paper proposes PFHI, an end-to-end frame-work designed to alleviate the accuracy–efficiency trade-off in joint monocular vehicle shape–pose reconstruction. First, we design a prior-factor-driven attention to effectively aggregate prior information, thereby improving shape reconstruction accuracy while notably reducing computational cost. Second, we introduce a hierarchical interactive attention module that decouples object–object and object–scene interactions, enhancing the structural representation for pose estimation. In addition, Dynamic Tanh (DyT) is incorporated as a lightweight normalization alternative to improve convergence stability and inference efficiency without relying on statistic-dependent normalization opera-tions. Experimental results on the ApolloCar3D dataset show that, compared to the existing state-of-the-art model, the proposed method improves the A3DP-Abs and A3DP-Rel indicators by 1.87% and 6.26%, respectively, while reducing inference time by 41.46%.
Monocular 3D vehicle shape and pose estimation is of great significance for scene understanding in autonomous driving. However, it encounters challenges owing to the lack of depth information, complex interactions between objects, and efficient modeling requirements. In this paper, we propose an end-to-end framework that integrates prior factors and hierarchical interaction to jointly model the 3D shape and 6 DOF pose of a vehicle. First, we design a prior-factor-driven attention to effectively aggregate prior information, thereby improving shape reconstruction accuracy while notably reducing computational cost. Second, we introduce a hierarchical interactive attention module that decouples object–object and object–scene interactions, enhancing the structural representation for pose estimation. We replace the traditional normalization operation with Dynamic Tanh (DyT) for the first time, further improving the model’s convergence stability and inference efficiency. Experimental results on the ApolloCar3D dataset show that, compared to the existing state-of-the-art model, the proposed method improves the AD3P-Abs and AD3P-Rel indicators by 1.87% and 6.26%, respectively, while reducing inference time by 41.46%.
With the rapid expansion of railway networks globally, ensuring rail infrastructure safety through efficient detection methods has become critical. Traditional inspection systems face limitations in flexibility, adaptability to adverse weather, and multifunctional integration. This study proposes a ground-air collaborative multi-source detection system that integrates 3D light detection and ranging (LiDAR)-based point cloud imaging and deep learning-driven intrusion detection. The system employs a lightweight rail inspection vehicle equipped with dual LiDARs and an Astro camera, synchronized with an unmanned aerial vehicle (UAV) carrying industrial-grade LiDAR. We propose an improved LiDAR odometry and mapping with sliding window (LOAM-SLAM) algorithm enables real-time dynamic mapping, while an optimized iterative closest point (ICP) algorithm achieves high-precision point cloud registration and colorization. For intrusion detection, a You Only Look Once version 3 (YOLOv3)-ResNet fusion model achieves a recall rate of 0.97 and precision of 0.99. The system’s innovative design and technical implementation offer significant improvements in railway track inspection efficiency and safety. This work establishes a new paradigm for adaptive railway maintenance in complex environments.
27 February 2026 in Measurement Science and Technology
Laser simultaneous localization and mapping (SLAM) systems commonly encounter challenges such as unstable feature extraction, significant attitude estimation errors, and limited global consistency, particularly in complex environments or when facing feature degradation. This study introduces a novel laser SLAM approach that leverages adaptive vector response feature extraction. By integrating point-to-sensor distance and the rate of change of surface normals, the method dynamically adjusts the feature extraction threshold to enhance the robustness and coherence of feature selection. A weighted residual mechanism is implemented to dynamically assign weights to matching pairs according to residual magnitude, directional consistency, and normal variation intensity. This enhances the precision and robustness of graph optimization. Moreover, a dynamic reference map is constructed in this study, and point cloud matching with the map in key frames is performed to serve as an absolute pose constraint, effectively reducing cumulative drift. Extensive experiments conducted on both the publicly available KITTI dataset and a proprietary dataset validate the efficacy of the proposed approach. The findings demonstrate superior performance of our system compared to existing methods in terms of accuracy and robustness, particularly in scenarios characterized by sparse structures, thus affirming its suitability for reliable real-world implementation.