Three-dimensional object detection using LiDAR has attracted significant attention due to its resilience to lighting conditions and ability to capture detailed geometric information. However, existing methods still face challenges, such as a high proportion of background points in the sampled point set and limited accuracy in detecting distant objects. To address these issues, we propose semantic-guided proposal sampling-RCNN (SPS-RCNN), a multi-stage detection framework based on point–voxel fusion. The framework comprises three components: a voxel-based region proposal network (RPN), a keypoint sampling stream (KSS), and a progressive refinement network (PRN). In the KSS, we propose a novel semantic-guided proposal sampling (SPS) method, which increases the proportion of foreground points and enhances sensitivity to outliers through multilevel sampling that integrates proposal-based local sampling and semantic-guided global sampling. In the PRN, a cascade attention module (CAM) is employed to aggregate features from multiple subnets, progressively refining region proposals to improve detection accuracy for medium- and long-range objects. Comprehensive experiments on the widely used KITTI dataset demonstrate that SPS-RCNN improves detection accuracy and exhibits enhanced robustness across categories compared to the baseline.

Given the harsh working conditions of underwater pan-tilt systems, their energy consumption management is particularly crucial. This study proposes an underwater pan-tilt operation method with an automatic wake-up mechanism, which activates only upon target detection, replacing conventional timer-based triggering. Furthermore, departing from fixed-duration observation strategies, we introduce a Q-learning algorithm to optimize operational modes. The algorithm dynamically adjusts working modes based on surrounding biological activity frequency: employing a low-power mode (reduced energy consumption with lower monitoring intensity) during periods of sparse biological presence and switching to a high-performance mode (extended observation duration, higher energy consumption, and enhanced monitoring intensity) during frequent biological activity. Simulation results demonstrate that compared to fixed-duration observation schemes, the proposed optimization strategy achieves a 11.11% improvement in monitoring effectiveness while achieving 16.21% energy savings.

Underwater towed survey systems are widely used for marine observation, resource exploration, and target identification. While high-speed towing is increasingly required to improve operational efficiency, conventional single-stage towed systems face a critical trade-off: active heave compensation systems are complex and costly, whereas purely passive configurations lack sufficient disturbance rejection at higher speeds. To address this gap, this study proposes a two-stage towing system consisting of a vessel, heavy cable, depressor, light cable, and detection towed body, where the depressor functions as a mechanical low-pass filter. The depressor reduces vessel-induced heave motion transmission by approximately 79% compared with a conventional single-stage system. CFD simulations are conducted to evaluate hydrodynamic performance and extract key coefficients. A lumped-mass dynamic model is established for time-domain motion simulations. An integral sliding-mode controller with vessel heave feedforward compensation is designed to enhance depth-tracking capability. The active controller eliminates step response overshoot and provides robust depth regulation under wave disturbances. Sea trials under real ocean conditions validate the system’s motion stability, demonstrating satisfactory depth-keeping performance at high towing speeds. The simulation results show good agreement with experimental data, confirming the effectiveness of the proposed system and dynamic model. This work offers a practically validated towing platform solution for high-precision underwater survey operations.

Screw propulsors have attracted increasing attention for their potential applications in amphibious vehicles and benthic robots, owing to their ability to perform both terrestrial and underwater locomotion. To investigate their hydrodynamic characteristics, a two-stage numerical analysis was carried out. In the first stage, steady-state simulations under various advance coefficients were conducted to evaluate the influence of key geometric parameters on propulsion performance. Based on these results, a representative configuration was then selected for transient analysis to capture unsteady flow features. In the second stage, a Detached Eddy Simulation approach was employed to capture unsteady flow features under three rotational speeds. The flow field information was analyzed, and the mechanisms of vortex generation, instability, and dissipation were comprehensively studied. The results reveal that the propulsion process is dominated by the formation and evolution of tip vortices, root vortices, and cylindrical wake vortices. As rotation speed increases, vortex structures exhibit a transition from ordered spiral wakes to chaotic turbulence, primarily driven by centrifugal instability and nonlinear vortex interactions. Vortex breakdown and energy dissipation are intensified downstream, especially under high-speed conditions, where vortex integrity is rapidly lost due to strong shear and radial expansion. This hydrodynamic behavior highlights the fundamental difference from conventional propellers, and these findings provide theoretical insight into the flow mechanisms of screw propulsion.

Underwater fish detection and multi-object tracking support marine ecological monitoring, underwater robot inspection, and fish behavior analysis. In a tracking-by-detection framework, this study targets three specific problems: unstable detection inputs for small- and medium-scale elongated fish, short-term trajectory breaks when low-confidence true detections are discarded, and identity switches caused by local candidate-edge competition when fish cross or move close together. We address these problems with a joint framework that combines the FishDet-XR detector and the FishBoT-SLR-TDR tracker. FishDet-XR stabilizes detector inputs through the proposed SFP-AFPN-XR feature-fusion neck, strip-shaped directional feature modeling, the Simple Parameter-Free Attention Module (SimAM), and positive-prior sampling. FishBoT-SLR-TDR improves tracker-side association through Spatially-Gated Low-Score Recovery (SLR) and Trajectory-Direction Reranking (TDR). Experiments on BrackishMOT-onlyfish show that the framework improves detection localization and tracking continuity while maintaining real-time inference. At the detection stage, FishDet-XR improves mean average precision at an Intersection over Union threshold of 0.50 (mAP₅₀) from 74.48% to 76.19%, mean average precision averaged over Intersection over Union thresholds from 0.50 to 0.95 (mAP_50–95) from 40.72% to 43.55%, and Precision from 86.75% to 88.00% compared with YOLO11s-640, while maintaining 54.76 frames per second (FPS). Compared with the YOLO11s-640 + BoT-SORT baseline, the final detection-tracking chain increases Higher Order Tracking Accuracy (HOTA) from 40.090 to 42.497 and identity F1 score (IDF1) from 51.372 to 54.926, while reducing identity switches (IDSW) from 184 to 159 and trajectory fragmentations (Frag) from 270 to 253.