Yanzhe Wang

Yanzhe Wang

Auto-generated profile, based on publicly available information

Yanzhe Wang

State Key Laboratory of Fluid Power and Mechatronic Systems, School of Mechanical Engineering, Zhejiang University, Hangzhou, China

Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

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

Representative Publications

View All Publications
14 December 2025 in Nature Communications
Verified
Journal Article
Open access
PDF file
YWYanzhe Wang
Yanzhe Wang
HGHaotian Guo
Haotian Guo
HWHao Wu
Hao Wu
HDHuixu Dong
Huixu Dong
Integrating human-level haptic perception into soft grippers promises safer robotic grasping and improved human–robot interaction. While visual-tactile sensors promise high perception resolution at low cost, they often sacrifice compliance to maintain optical stability, hindering non-planar contact perception. We present FlexiRay, a soft gripper that integrates visual-tactile sensing with the Fin Ray Effect to achieve high compliance, broad sensory coverage, and multimodal capability. Combining a multi-layered flexible substrate, an optimized multi-mirror optical system, and a decoupled deep learning framework, FlexiRay replicates five of the seven human tactile modalities, including force, contact location, texture, temperature, and proprioception, with a single camera. It achieves 0.17 N force accuracy, 0.96 mm spatial resolution, 0.24 mm proprioception accuracy, and 1.17 °C temperature accuracy while maintaining over 90% effective coverage. FlexiRay empowers compliant grasping, safe collaboration, and intelligent teleoperation, underscoring its potential to propel service robotics toward enhanced intelligence, safety, and real-world utility.

24 June 2024 in Animal Microbiome
Verified
Journal Article
Open access
PDF file
DDDewei Du
Dewei Du
YWYanzhe Wang
Yanzhe Wang
YGYongji Gao
Yongji Gao
LFLei Feng
Lei Feng
Ziye ZhangZiye Zhang
Ziye Zhang
ZHZhiyong Hu
Zhiyong Hu
Background: The rumen is a crucial digestive organ for dairy cows. The rumen microbiota assists in the digestion of plant feed through microbe-mediated fermentation, during which the plant feed is transformed into nutrients for the cow's use. Variations in the composition and function of the rumen microbiome affect the energy utilization efficiency of dairy cows, which is one of the reasons for the varying body condition scores (BCSs). This study focused on prepartum Holstein dairy cows to analyze differences in rumen microbiota and metabolites among cows with different BCSs. Twelve prepartum dairy cows were divided into two groups, low BCS (LBCS, BCS = 2.75, n = 6) and high BCS (HBCS, BCS = 3.5, n = 6), to explore differences in microbial composition and metabolites. Results: In the HBCS group, the genera within the phylum Firmicutes exhibited stronger correlations and greater abundances. Phyla such as Firmicutes, Patescibacteria, Acidobacteriota, Euryarchaeota, and Desulfobacterota, in addition to most of their constituent microbial groups, were significantly more abundant in the HBCS group than in the LBCS group. At the genus level, the abundances of Anaerovibrio, Veillonellaceae_UCG_001, Ruminococcus_gauvreauii_group, Blautia, Eubacterium, Prevotellaceae_YAB2003_group, Schwartzia, and Halomonas significantly increased in the HBCS group. The citrate cycle, involved in carbohydrate metabolism, exhibited a significant enrichment trend, with a notable increase in the abundance of its key substrate, citrate, in the HBCS group. This increase was significantly positively correlated with the differential bacterial genera. Conclusion: In this study, prepartum dairy cows with higher BCS exhibited greater abundance of Firmicutes. This study provides theoretical support for microbiological research on dairy cows with different BCSs and suggests that regulating the rumen microbiome could help maintain prepartum dairy cows within an optimal BCS range.

13 September 2023 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
LFLei Fu
Lei Fu
ZMzepeng ma
zepeng ma
LZLibin Zhang
Libin Zhang
YWYanzhe Wang
Yanzhe Wang
Developing fault diagnosis algorithms presents a significant challenge for rotary bearing signals with composite defects due to their inherent characteristics of non-linearity, time-variability, instability, and uncertainty. Hence, this paper proposes a novel diagnostic architecture, IRTCog, based on variable visual-angle infrared thermography (V-IRT) images and an asymmetric convolutional neural network, so as to overcome the insufficient samples, excessive parameters, and overfitting. V-IRT images that adequately characterize composite defects are considered for model training. Besides, hybrid activation functions and asymmetric convolution processes are designed to improve the accuracy and efficiency of the diagnostic model without increasing the parameter count. Finally, transfer learning is introduced to reduce model dependence on sample size and training time. The experimental results demonstrate that the proposed method reduces the training time by 72.9% and the diagnosis accuracy is close to 99%, indicating its superiority compared to other mainstream models.

9 September 2024 in Complex & Intelligent Systems
Verified
Journal Article
Open access
PDF file
YWYanzhe Wang
Yanzhe Wang
QYQian Yang
Qian Yang
WQWeiwei Qu
Weiwei Qu
In Automated Fiber Placement (AFP), the substantial structure of the placement robot, the variable mold shapes, and the limited free space pose significant challenges for planning collision-free robot transitions. The task involves planning a collision-free path within the robot's high-dimensional configuration space. Informed RRT* is a common approach for such problems but often struggles with efficiency and path quality in environments with large informed sampling spaces influenced by obstacles. To address these issues, this paper proposes an improved Informed RRT* algorithm with a Local Knowledge Acceleration sampling strategy (LKA-Informed RRT*), aimed at enhancing planning efficiency and adaptability in complex obstacle settings. An Adaptive Sampling Control (ASC) rate is introduced, measuring the algorithm’s convergence speed, guides the algorithm to switch between informed and local sampling adaptively. The proposed local sampling method uses failure nodes from the exploration process to estimate obstacle distributions, steering sampling toward regions that expedite path convergence. Experimental results show that LKA-Informed RRT* significantly outperforms state-of-the-art algorithms in convergence efficiency and path cost. Compared to the original Informed RRT*, the proposed method reduces planning time by about 60%, substantially boosting efficiency for collision-free transitions in complex environments.

25 November 2025 in IEEE Transactions on Automation Science and Engineering
Verified
Journal Article
Closed access
YWYanzhe Wang
Yanzhe Wang
WYWei Yu
Wei Yu
HWHao Wu
Hao Wu
HGHaotian Guo
Haotian Guo
HDHuixu Dong
Huixu Dong
We propose a novel framework, SA-DEM, for extrinsic non-grasping manipulation of ungraspable objects in robotics. This approach is grounded in dual-stage reinforcement learning and decouples the overall task into two sequential phases: interactive mode decision-making and manipulation action planning. Notably, this framework innovatively incorporates the stiffness information of manipulated objects into the decision-making process, enabling the robot to autonomously perceive, decide, and plan manipulation strategies for objects with diverse physical attributes. The first phase of SA-DEM involves a high-level agent responsible for planning the grasping pose of objects and their interaction locations with the environment, based on the initial state of the objects, observations from environmental point clouds, stiffness representations, and prior knowledge of grasping regions. The second phase is executed by a low-level agent, which focuses on planning specific manipulation actions such as poking and flipping. These actions are derived from autonomous exploration during the training process, negating the need for manual customization. Both agents employ a hybrid discrete-continuous action space along with time-abstracted and spatially grounded representations centered around the point cloud, culminating in a unified actor-critic reinforcement learning framework.

1 June 2019
Verified
Conference Paper
Closed access
YTYe Tao
Ye Tao
YWYanzhe Wang
Yanzhe Wang
CSCao Shi
Cao Shi
XWXiaodong Wang
Xiaodong Wang
CXCanhui Xu
Canhui Xu
ZXZhifang Xu
Zhifang Xu
This paper presents a user profile construction approach by analyzing user action log data obtained from cross-domain platforms. The concept of Semantic Activity Space was proposed to represent user characteristics from multiple dimensions, where the basic attributes and values are calculated and integrated from the logs. Non-parametric methods are used to classify and visualize the differences in level and type of activity preferences of users. Log items are integrated from heterogeneous modules of information management systems in manufacturing enterprises. The experiment results indicate that by extracting and analyzing key features from semantic activity space, the characters of a typical user can be properly profiled.