Chunyun Fu

The majority of existing light detection and ranging (LiDAR)‐based odometry solutions rely on rudimentary geometric features, such as points, lines, or planes, that fall short of fully capturing the intricate characteristics of surrounding environments. In this study, we introduce an innovative landmark‐based LiDAR odometry (LO) method for structured environments with rich landmark information. This odometry harnesses advanced feature extraction techniques to effectively exploit the holistic exterior attributes of environmental landmarks. Our focus is on augmenting pose estimation for robotic systems. The vehicle pose estimation is achieved through a two‐stage sequential process, that is, a horizontal pose estimation stage followed by a vertical pose estimation step. To facilitate effective landmark registration, we propose a comprehensive index that quantifies the degree of similarity between landmarks. This index meticulously considers two pivotal aspects of landmarks: dimension and shape. To evaluate the efficacy of our proposed algorithm, we employ the widely recognized KITTI dataset, along with experimental data collected by an unmanned ground vehicle platform. Both qualitative and quantitative analyses demonstrate the effectiveness of our algorithm compared with state‐of‐the‐art LO solutions. Quantitatively, our method achieves an average translational error of 1.59% and an average rotational error of 0.57°/100 m on the KITTI data set, representing improvements of 17.19% and 21.92%, respectively, over the best competing method (G‐ICP). Furthermore, the proposed method possesses the capability to concurrently construct maps with instance‐level landmark separation.

To address the high computational complexity and data association challenges in multi-sensor networks, this paper proposes a spatial-adaptive confidence-weighted Poisson multi-Bernoulli mixture (SACW-PMBM) algorithm that applies a divide-and-conquer fusion strategy, reformulating the global multi-target estimation problem into a series of localized, weighted tasks. Existing distributed PMBM fusion methods, such as those employing the weighted arithmetic average (WAA) or generalized covariance intersection (GCI), utilize a single scalar weight per sensor, which fails to capture spatially varying information quality. The primary contribution of this work is to replace these scalar weights with a state-dependent confidence function, ω^(s) (x), that dynamically models sensor reliability across the state space. This confidence function is constructed using a variational autoencoder (VAE) that jointly represents local information entropy and environmental disturbances. We subsequently derive a novel fusion criterion by minimizing the confidence-weighted Kullback-Leibler (KL) divergence between the local sensor posteriors and the fused global density. A computationally tractable hierarchical architecture is also developed to manage hypothesis complexity. Simulation results in challenging scenarios with severe occlusion and non-uniform clutter demonstrate that the proposed SACW-PMBM algorithm achieves superior tracking accuracy and robustness compared to benchmark centralized PMBM (C-PMBM), WAA, and GCI fusion methods.

This systematic review critically examines the benefits and challenges of high-compression-ratio (CR) implementation in hybrid-dedicated engines, recognizing CR increase as a pivotal strategy for enhancing the indicated thermal efficiency to achieve carbon peak and carbon neutrality goals. However, excessively high CRs face critical constraints, including intensified knock propensity, increased heat transfer (HTR) losses, reduced combustion stability, augmented dissociation losses, and cold-start misfire risks. The feasibility and necessity of CR enhancement in hybrid systems were comprehensively evaluated based on these factors, with fundamental mechanisms of the detrimental effects elucidated. To address these challenges, optimized countermeasures were synthesized: knock suppression via high-octane fuels, EGR technology, lean combustion, and in-cylinder water injection; heat transfer reduction through thermal barrier coatings and independent CR/expansion-ratio control; misfire risk monitoring using ion current or cylinder pressure sensors. These approaches provide viable pathways to overcome high-CR limitations and optimize engine performance. Nevertheless, current research remains confined to isolated solutions, warranting future focus on integrated optimization mechanisms investigating synergistic interactions of multiple strategies under high-CR conditions.

This paper proposes a semantic SLAM methodology for autonomous driving, integrating panoramic camera and LiDAR fusion to address pose estimation inaccuracies and lack of semantic information in dynamic environments. On the LiDAR side, panoramic semantic segmentation results are projected onto point clouds and combined with clustering for instance segmentation, classifying points into unknown-motion landmarks, static landmarks, and ground points. Precise pose estimation is then achieved using purely static and semi-static landmarks. The visual component incorporates data augmentation and instance segmentation of front-view images, leveraging SuperPoint for feature extraction. 3D dynamic landmarks are projected onto segmentation results to generate dynamic-static masks, enabling the elimination of feature points within dynamic regions. A robust visual odometry system is implemented through the ORB-SLAM2 tracking framework, enhanced by SuperGlue feature association within a multi-level tracking mechanism. At the fusion level, a decision-layer fusion algorithm integrates both odometry sources. Furthermore, pose optimization is achieved through dual loop closure detection with inputs from both LiDAR and visual sources. The proposed system was validated using the CQU01-CQU04 datasets collected on a UGV, with high-precision GNSS serving as the ground truth. Experimental results demonstrate that the system achieves mean of average error of estimated positions, standard deviation of estimated position errors, and position error at the trajectory end of 0.446m, 0.236m, and 0.775m, respectively. With loop closure enabled, the performance metrics showed enhancement to 0.320m, 0.181m, and 0.617m. This research demonstrates that our method improves localization accuracy while constructing semantic map comprising purely static elements.

A novel sliding mode control (SMC) strategy incorporating an adaptive super-twisting algorithm is developed for permanent magnet synchronous motors (PMSMs), effectively mitigating high-frequency chattering while enhancing external disturbance rejection capabilities. Initially, a sliding surface is crafted based on the dynamic characteristics of the PMSM and real-time feedback. The super-twisting algorithm is subsequently applied adaptively to dynamically adjust the control effort required to maintain the sliding mode state, thereby ensuring precise and prompt intervention to uphold system stability and enhance response speed. Additionally, in light of operational challenges such as road-induced load disturbances, a Lyapunov-based disturbance observer is proposed for precise load torque estimation in PMSM systems. The efficacy of the proposed control and observation methods is substantiated through a hardware-in-the-loop experiment test, demonstrating that the developed sliding mode controller, leveraging the adaptive super-twisting algorithm, exhibits superior tracking and disturbance rejection capabilities, reduces steady-state current error, and bolsters system parameter robustness, and the modified extended state observer (MESO) exhibits commendable estimation performance.

Local trajectory planning (TP) for collision avoidance typically comprises path planning (PP) and velocity planning (VP). Various objectives must be fulfilled in a PP task, and the majority of prior works integrate all objectives into a unified cost function. To prioritize the dominant objectives of each PP stage, we propose to decouple the PP task into two separate subtasks, enabling the logical establishment of subtask-oriented segmented PP methods. First, based on risk evaluation of four vehicle vertices, an improved artificial potential field was established. Second, a novel transit point selection method was applied to decouple the PP task into two segmented subtasks. Then, the optimization problem was converted into a multi-attribute decision-making (MADM) problem and the technique for order preference by similarity to ideal solution (TOPSIS) method was utilized to obtain two optimal segmented paths. Finally, a velocity planner based on cubic polynomial, in conjunction with a support vector machine-based stability classifier, was designed. The proposed trajectory planner was then verified in six typical driving scenarios, including both simulation and real vehicle studies. Verification results demonstrate that the proposed planner effectively decouples the PP task and achieves a safe, comfortable, efficient, and trackable trajectory.

Current camera-based tracking-by-detection methodologies in the literature overlook the significance of detection box localization confidence, generally assuming that objects with low detection confidence are highly occluded and, therefore, either ignored or deprioritized in the matching process. Furthermore, appearance similarity is typically neglected when matching these low-confidence objects. This oversight presents a critical research gap as objects with low detection confidence might still exhibit a clear appearance, and those with high detection confidence can present imprecise localization or ambiguous appearance, factors not accounted for in existing methods. To address this gap, our work contributes a novel framework, Localization-Guided Track (LG-Track), which, for the first time, integrates localization confidence of camera detections in multi-object tracking (MOT). LG-Track takes into account both appearance clarity and localization precision of detection boxes, incorporating a novel deep association mechanism that enhances tracking performance. Based on the localization and classification confidence of detection boxes, different cost matrices are employed in different levels of the proposed deep association mechanism to achieve enhanced matching accuracy. Our method, validated through rigorous experimentation on the MOT17 and MOT20 benchmarks, demonstrates superior performance over current compared state-of-the-art tracking methods. Committed to furthering research in this field, we have made our code accessible to the community at https://github.com/mengting2023/LG-Track .

In the classical tracking-by-detection (TBD) paradigm, detection and tracking are separately and sequentially conducted, and data association must be properly performed to achieve satisfactory tracking performance. In this paper, a new multi-object tracking framework is proposed, which integrates object detection and multi-object tracking into a single model. The proposed tracking framework eliminates the complex data association process in the classical TBD paradigm, and requires no additional training. Secondly, the regression confidence of historical trajectories is investigated, and the possible states of a trajectory (weak object or strong object) in the current frame are predicted. Then, a confidence fusion module is designed to guide non-maximum suppression for trajectories and detections to achieve ordered and robust tracking. Thirdly, by integrating historical trajectory features, the regression performance of the detector is enhanced, which better reflects the occlusion and disappearance patterns of objects in real world. Lastly, extensive experiments are conducted on the commonly used KITTI and Waymo datasets. The results show that the proposed framework can achieve robust tracking by using only a 2D detector and a 3D detector, and it is proven more accurate than many of the state-of-the-art TBD-based multi-modal tracking methods. The source codes of the proposed method are available at https://github.com/wangxiyang2022/MMF-JDT .