Wenkai Zhu

Wenkai Zhu

Auto-generated profile, based on publicly available information

Wenkai Zhu

Southeast University

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
23 October 2025 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
BWBenwu Wang
Benwu Wang
Xu LiXu Li
Xu Li
QXQimin Xu
Qimin Xu
WZWenkai Zhu
Wenkai Zhu
YDYinan Du
Yinan Du
BLBaidan Li
Baidan Li
Recently, LiDAR-based vehicle-infrastructure cooperative (VIC) perception has shown an advantage in expanding the horizon of connected autonomous vehicles (CAVs), enabling occlusion-aware 3-D scene understanding. Limited communication bandwidth, however, hampers multi-agent cooperation in the urban Internet of Things (IoT) environments. Existing solutions often compress or implicitly filter high-resolution features, resulting in semantic redundancy or information loss, which degrades the overall performance. To tackle this, we propose VICooper, a communication-efficient VIC perception framework. Driven by the stable context of static infrastructure LiDARs, VICooper employs offline background mapping (BgM) to extract foreground points of interest, thereby offering explicit guidance for communication reduction. For the sparse yet critical foreground point clouds, we introduce the multidimensional foreground backbone (MdFB) that incorporates geometric cues, including height, scale, and spatial density, to enrich feature encoding. To address the fusion imbalance between dense vehicle-side and sparse roadside features, we customize a progressive bidirectional feature aggregation (PbFA) using a deformable transformer to capture inter-agent mutual correlations, thereby enabling the deep coupling of heterogeneous agents under asymmetric information. Extensive evaluations on real-world VIC benchmark validate that VICooper achieves superior performance with substantially lower bandwidth, demonstrating its potential in intelligent transportation IoT ecosystems.

14 July 2025 in IEEE Sensors Journal
Verified
Journal Article
Closed access
KWKun Wei
Kun Wei
Xu LiXu Li
Xu Li
WZWenkai Zhu
Wenkai Zhu
WHWeiming Hu
Weiming Hu
YHYue Hu
Yue Hu
HCHaoyang Che
Haoyang Che
Learning-based multi-sensor fusion offers a promising solution for the autonomous vehicle localization in GNSS-denied environments. However, existing methods primarily focus on the real-time pose estimation while neglecting the suppression of accumulated errors in historical poses. To address this problem, we propose a multi-stage learning-based fusion framework that integrates pose transformation estimation, local pose correction, and global pose optimization. In Stage1, we deeply fuse geometric structure data from LiDAR, color texture data from camera, and motion state data from IMU to accurately estimate the pose transformation. Continuous transformations are accumulated to generate a preliminary pose sequence. Entering Stage2, we constantly introduce absolute pose data from UWB to rapidly correct Stage1 poses in the local region. Accumulated errors in Stage1 poses are evaluated in this region by comparing poses before and after the UWB correction. Reaching Stage3, we fully leverage the spatio-temporal correlation of accumulated errors to globally optimize Stage1 poses. Measured accumulated errors in the UWB-covered region are employed as the benchmark to recursively predict accumulated errors in the non-covered region, generating a more accurate pose sequence. We conduct qualitative and quantitative experiments on KITTI dataset and our dataset to compare our fusion framework with state-of-the-art methods. Experimental results show the competitiveness of our fusion framework in improving the localization performance in GNSS-denied environments.

9 February 2026 in IEEE Robotics and Automation Letters
Verified
Journal Article
Closed access
ZWZihang Wang
Zihang Wang
XLXu Li
Xu Li
GZGuanyu Zong
Guanyu Zong
XCXieyuanli Chen
Xieyuanli Chen
DKDong Kong
Dong Kong
KWKaiyi Wang
Kaiyi Wang
Place recognition (PR) is a key component of simultaneous localization and mapping (SLAM) in autonomous vehicles and robotics. By efficiently matching descriptors generated from the current scene with a prebuilt reference database, existing PR methods enable accurate vehicle re-localization. However, achieving robust and efficient PR remains challenging, particularly in scenarios with sensor corruptions, such as camera overexposure and LiDAR beam missing, which degrade the performance of traditional PR systems. To address this issue, we propose LiteFT-PR, a fault-tolerant and lightweight LiDAR-camera fusion PR network. To balance accuracy and efficiency under extreme conditions, we adopt a knowledge distillation framework for PR tasks: the teacher network learns robust representations under corrupted sensing, while the student distills and inherits this knowledge for lightweight deployment. In addition, we design efficient multimodal fusion mechanisms for both networks, enabling robust place recognition under various sensor corruption scenarios while significantly reducing computational overhead. To meet real-world deployment needs, we build a dataset covering multiple types of sensor failures and conduct extensive field experiments. Experimental results demonstrate that LiteFT-PR, with only 7.24 M of parameters, outperforms state-of-the-art PR methods in both accuracy and efficiency.

10 November 2025 in IEEE Sensors Journal
Verified
Journal Article
Closed access
ZWZihang Wang
Zihang Wang
Xu LiXu Li
Xu Li
CTChi Tian
Chi Tian
LXLiang Xia
Liang Xia
WZWenkai Zhu
Wenkai Zhu
YPYiming Peng
Yiming Peng
In dense maritime environments with complex shipping routes, small-scale vessel radar imaging, frequent occlusions, and severe sea clutter present catastrophic challenges for maritime radar-based multi-vessel tracking. Precisely modeling vessel targets under such interference and achieving robust tracking through occluded routes is therefore critical for maritime surveillance tasks. To this end, we propose RadarTracker, a novel end-to-end temporal Transformer framework for X-band marine radar multi-vessel tracking. RadarTracker consists of a CNN-based feature extractor, a density enhancement encoder, and an occlusion-aware dynamic decoder. Specifically, the encoder is built upon a Deformable Transformer backbone, integrated with a density-aware module and density feature enhancement to address the low signal-to-noise ratio and densely distributed vessel targets in radar imagery. Subsequently, to achieve accurate trajectory regression under varying traffic densities and complex occlusions along shipping routes, we design an occlusion-aware decoder with a dynamic query allocation mechanism that enables seamless target initialization, identity preservation, and short-term re-identification, effectively balancing tracking accuracy and efficiency under diverse traffic conditions. Finally, we construct a high-density shipping lane dataset collected along the Yangtze River in China to validate the proposed approach. Experimental results demonstrate that RadarTracker achieves a consistent 5% performance improvement over state-of-the-art baselines across varying traffic densities, while maintaining deployment-friendly computational efficiency. The implementation of our method will be publicly available at https://github.com/wangzihanggg/RadarTracker.

24 January 2026 in Expert Systems with Applications
Verified
Journal Article
Closed access
PNpeizhou ni
peizhou ni
WZWenkai Zhu
Wenkai Zhu
Feng JiangFeng Jiang
Feng Jiang
BWBenwu Wang
Benwu Wang
YHYue Hu
Yue Hu

1 January 2026 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
HWHaitao Wang
Haitao Wang
Xu LiXu Li
Xu Li
QXQimin Xu
Qimin Xu
BWBenwu Wang
Benwu Wang
WZWenkai Zhu
Wenkai Zhu
ZWZihang Wang
Zihang Wang