Yanke Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yanke Wang has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
- Markus Reischl
Institute for Automation and Applied Informatics (IAI), Karlsruhe Institute of Technology (KIT), Eggenstein-Leopoldshafen, Germany
4 shared publications - Christian Pylatiuk
Institute for Automation and Applied Informatics (IAI) Karlsruhe Institute of Technology (KIT) Karlsruhe Germany
3 shared publications - Ralf Mikut
Institute for Automation and Applied Informatics (IAI), Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany
3 shared publications Jiangtao Wen
2 shared publications- Ravindra Peravali
Karslruhe Institute of Technology KIT Karlsruhe Germany
2 shared publications - Yuxing Han
Shenzhen International Graduate School Tsinghua University
2 shared publications
Latest Publications
- 19 June 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 349
Multi-scale semantics meet SE(3) geometry: A sequence-aware hierarchical VLA for long-horizon manipulation
- 19 June 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 349
Multi-scale semantics meet SE(3) geometry: A sequence-aware hierarchical VLA for long-horizon manipulation
- Open Access
- 23 April 2026
- Published by Elsevier in SSRN Electronic Journal
Multi-scale Semantics Meet SE(3) Geometry: A Sequence-Aware Hierarchical VLA for Long-Horizon Manipulation
Robotic manipulation in unstructured environments, such as construction sites, requires agents to simultaneously balance task-level planning and millimeter-level geometric precision. However, existing Vision-Language-Action (VLA) models often struggle to handle these distinct scales effectively, and geometric diffusion models typically lack the semantic guidance necessary for sequential decision-making. To address the challenges, we propose the Sequence-Aware Hierarchical VLA (SA-HVLA), a unified generative framework designed for high-precision sequential tasks, taking multi-mode rebar tying as a stringent real-world testbed. Our method introduces two key contributions: (i) a hierarchical language grounding mechanism that decouples textual instructions into global task directives and local node specifications; and (ii) a history-conditioned $SE(3)$-equivariant diffusion backbone that seamlessly integrates sequence-aware semantic features via invariant modulation, ensuring geometric consistency across long horizons and reducing demonstration data requirement. Experiments in both simulation setup and real-world application demonstrate that our hierarchical architecture reduces geodesic pose error by approximately 70\% compared with single-resolution baselines in noisy physical environments. Additionally, the $SE(3)$-equivariant formulation maintains strong robustness, exhibiting only 7.1\% variance under significant viewpoint changes. Furthermore, the model effectively mitigates error accumulation in multi-node sequential tying task, maintaining high success rates where memoryless baselines fail.
- 23 April 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessMulti-scale Semantics Meet SE(3) Geometry: A Sequence-Aware Hierarchical VLA for Long-Horizon Manipulation
Robotic manipulation in unstructured environments, such as construction sites, requires agents to simultaneously balance task-level planning and millimeter-level geometric precision. However, existing Vision-Language-Action (VLA) models often struggle to handle these distinct scales effectively, and geometric diffusion models typically lack the semantic guidance necessary for sequential decision-making. To address the challenges, we propose the Sequence-Aware Hierarchical VLA (SA-HVLA), a unified generative framework designed for high-precision sequential tasks, taking multi-mode rebar tying as a stringent real-world testbed. Our method introduces two key contributions: (i) a hierarchical language grounding mechanism that decouples textual instructions into global task directives and local node specifications; and (ii) a history-conditioned $SE(3)$-equivariant diffusion backbone that seamlessly integrates sequence-aware semantic features via invariant modulation, ensuring geometric consistency across long horizons and reducing demonstration data requirement. Experiments in both simulation setup and real-world application demonstrate that our hierarchical architecture reduces geodesic pose error by approximately 70\% compared with single-resolution baselines in noisy physical environments. Additionally, the $SE(3)$-equivariant formulation maintains strong robustness, exhibiting only 7.1\% variance under significant viewpoint changes. Furthermore, the model effectively mitigates error accumulation in multi-node sequential tying task, maintaining high success rates where memoryless baselines fail.
- Open Access
Bird's-eye view safety monitoring for the construction top under the tower crane
The tower crane is involving more automated and intelligent operation procedure, and importantly, the application of automation technologies to the safety issues is imperative ahead of the utilization of any other advances. Among diverse risk management tasks on site, it is essential to protect the human workers on the workspace between the tower crane and constructed building top area (construction top) from the bird's-eye view, especially with Modular Integrated Construction (MiC) lifted. Also, the camera and Light Detection And Ranging (LiDAR) can capture abundant 3D information on site, which is however yet made the best use. Considering the safety protection for humans and tower cranes, we present an AI-based fully automated safety monitoring system for tower crane lifting from the bird's-eye view, surveilling to shield the human workers on the construction top and avoid cranes' collision by alarming the crane operator. The system achieved a 3D data fusion for localization of humans and MiCs by integrating the captured information from camera and LiDAR. The state-of-the-art methods were explored and implemented into our proposed software pipeline coupled with the hardware and display systems. Furthermore, we conducted an analysis of the components in the pipeline to verify the accuracy and effectiveness of the involved methods. The display and visualization on the real site proved that our system can serve as a valuable safety monitoring toolkit on site.
Open AccessBird's-eye view safety monitoring for the construction top under the tower crane
The tower crane is involving more automated and intelligent operation procedure, and importantly, the application of automation technologies to the safety issues is imperative ahead of the utilization of any other advances. Among diverse risk management tasks on site, it is essential to protect the human workers on the workspace between the tower crane and constructed building top area (construction top) from the bird's-eye view, especially with Modular Integrated Construction (MiC) lifted. Also, the camera and Light Detection And Ranging (LiDAR) can capture abundant 3D information on site, which is however yet made the best use. Considering the safety protection for humans and tower cranes, we present an AI-based fully automated safety monitoring system for tower crane lifting from the bird's-eye view, surveilling to shield the human workers on the construction top and avoid cranes' collision by alarming the crane operator. The system achieved a 3D data fusion for localization of humans and MiCs by integrating the captured information from camera and LiDAR. The state-of-the-art methods were explored and implemented into our proposed software pipeline coupled with the hardware and display systems. Furthermore, we conducted an analysis of the components in the pipeline to verify the accuracy and effectiveness of the involved methods. The display and visualization on the real site proved that our system can serve as a valuable safety monitoring toolkit on site.
- 9 October 2024
- Published by Springer Nature
- p. 22-32
Energy-Based Prior Latent Space Diffusion Model for Reconstruction of Lumbar Vertebrae from Thick Slice MRI
- 9 October 2024
- Published by Springer Nature
- p. 22-32
Energy-Based Prior Latent Space Diffusion Model for Reconstruction of Lumbar Vertebrae from Thick Slice MRI
- Open Access
- 1 October 2023
- Published by Walter de Gruyter GmbH in at - Automatisierungstechnik
- Vol. 71 (10) , 845-852
A fully automated touch-response behavior inspection pipeline on zebrafish larvae
A touch-evoked response of zebrafish larvae provides information on the mechanism of the gene functional expressions. Recently, an automated system has been developed for precise and repeated touch-response experimentation with minor human intervention. To quantify the collected data, we propose a fully automated multi-larvae touch-response behavior inspection pipeline based on larva tracking and segmentation. Experimental data with different treatments is analyzed by using the proposed inspection platform for demonstration, and the result proves that this platform can generate comparable touch-response behavior inspection readouts efficiently and automatically. The initial results were published in 31. Workshop Computational Intelligence, and this paper summarizes and extends the main work of the respective article.
- 1 October 2023
- Published by Walter de Gruyter GmbH in at - Automatisierungstechnik
- Vol. 71 (10) , 845-852
Open AccessA fully automated touch-response behavior inspection pipeline on zebrafish larvae
A touch-evoked response of zebrafish larvae provides information on the mechanism of the gene functional expressions. Recently, an automated system has been developed for precise and repeated touch-response experimentation with minor human intervention. To quantify the collected data, we propose a fully automated multi-larvae touch-response behavior inspection pipeline based on larva tracking and segmentation. Experimental data with different treatments is analyzed by using the proposed inspection platform for demonstration, and the result proves that this platform can generate comparable touch-response behavior inspection readouts efficiently and automatically. The initial results were published in 31. Workshop Computational Intelligence, and this paper summarizes and extends the main work of the respective article.
- Top Cited PapersOpen Access
- 17 April 2023
- Published by Springer Nature in Nature Communications
- Vol. 14 (1) , 1-10
Random fractal-enabled physical unclonable functions with dynamic AI authentication
A physical unclonable function (PUF) is a foundation of anti-counterfeiting processes due to its inherent uniqueness. However, the self-limitation of conventional graphical/spectral PUFs in materials often makes it difficult to have both high code flexibility and high environmental stability in practice. In this study, we propose a universal, fractal-guided film annealing strategy to realize the random Au network-based PUFs that can be designed on demand in complexity, enabling the tags’ intrinsic uniqueness and stability. A dynamic deep learning-based authentication system with an expandable database is built to identify and trace the PUFs, achieving an efficient and reliable authentication with 0% “false positives”. Based on the roughening-enabled plasmonic network platform, Raman-based chemical encoding is conceptionally demonstrated, showing the potential for improvements in security. The configurable tags in mass production can serve as competitive PUF carriers for high-level anti-counterfeiting and data encryption.
- 17 April 2023
- Published by Springer Nature in Nature Communications
- Vol. 14 (1) , 1-10
Top Cited PapersOpen AccessRandom fractal-enabled physical unclonable functions with dynamic AI authentication
A physical unclonable function (PUF) is a foundation of anti-counterfeiting processes due to its inherent uniqueness. However, the self-limitation of conventional graphical/spectral PUFs in materials often makes it difficult to have both high code flexibility and high environmental stability in practice. In this study, we propose a universal, fractal-guided film annealing strategy to realize the random Au network-based PUFs that can be designed on demand in complexity, enabling the tags’ intrinsic uniqueness and stability. A dynamic deep learning-based authentication system with an expandable database is built to identify and trace the PUFs, achieving an efficient and reliable authentication with 0% “false positives”. Based on the roughening-enabled plasmonic network platform, Raman-based chemical encoding is conceptionally demonstrated, showing the potential for improvements in security. The configurable tags in mass production can serve as competitive PUF carriers for high-level anti-counterfeiting and data encryption.
- 13 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 7 (2) , 858-865
High-Throughput Data Acquisition Platform for Multi-Larvae Touch-Response Behavior Screening of Zebrafish
Touch response of zebrafish larvae refers to the behaviors after the mechanical stimuli that varies under different conditions (different drugs), and statistical analysis of the behaviors in the organism level is becoming more attractive in drug screening. Collecting repeatable and reliable experimental data on the alive animals in a high-throughput manner is of importance and challenging. Thus, we for the first time introduce a multi-larvae experimentation platform for the fully automated data acquisition to screen the touch-response behaviors of zebrafish in the high-throughput manner. A multi-larvae touching strategy is designed for this platform, which provides a new alternative method to generate touch-response experimental data in a higher efficiency than in the single-larva case (in average 3.58 videos collected per well instead of only one, with 33.00 s duration on each larva instead of 55.95 s). As well, for the proof of the concept, the unknown parameters used for the strategy are decided by experiments. This platform will help in the high-throughput screening of the drugs in the whole animal (organism) level.
- 13 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 7 (2) , 858-865
High-Throughput Data Acquisition Platform for Multi-Larvae Touch-Response Behavior Screening of Zebrafish
Touch response of zebrafish larvae refers to the behaviors after the mechanical stimuli that varies under different conditions (different drugs), and statistical analysis of the behaviors in the organism level is becoming more attractive in drug screening. Collecting repeatable and reliable experimental data on the alive animals in a high-throughput manner is of importance and challenging. Thus, we for the first time introduce a multi-larvae experimentation platform for the fully automated data acquisition to screen the touch-response behaviors of zebrafish in the high-throughput manner. A multi-larvae touching strategy is designed for this platform, which provides a new alternative method to generate touch-response experimental data in a higher efficiency than in the single-larva case (in average 3.58 videos collected per well instead of only one, with 33.00 s duration on each larva instead of 55.95 s). As well, for the proof of the concept, the unknown parameters used for the strategy are decided by experiments. This platform will help in the high-throughput screening of the drugs in the whole animal (organism) level.
- 24 August 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 19 (4) , 3007-3019
An Automated Experimentation System for the Touch-Response Quantification of Zebrafish Larvae
Touch-response experimentation in zebrafish helps researchers better understand the link between genetics, drug effects, and behaviors. However, commonly manually conducted experimentation cannot fulfill a high-throughput screening and often delivers low accuracy and lacks reproducibility. Thus, the main aim of this work is to establish a fully automated robot-assisted experimentation system with minimal human participation to conduct the touch-response experimentation with freely swimming zebrafish larvae. Our designed system is able to undertake the role of repeated touch-response experiments at predefined specific location of the larvae in different ages and under different conditions, with high accuracy, robustness, and repeatability, and can also get comparable experimental results. The errors of the detection methods are less than 3 pixels and the offset errors of the touching points are less than 5%. Designed for high-efficiency experimentation, this system will promisingly release a great amount of the burden for the biological operators from touch-response experiments and may also have potential applications in other organisms for touch-evoked response analysis. Note to Practitioners—This article presented an automated touch-response experimentation system on zebrafish larvae, which can release a huge burden of the biological operators by achieving accurate, efficient, and repeated touch-response experiments. Manually conducted experiments are time-consuming and has low accuracy as well as difficult to do the quantification of the behaviors of the larvae. Our proposed system introduced an automated experiment pipeline and can also generate the behavior quantification of the larvae automatically, as the verification experiments confirmed. Basically, the research on the behaviors of zebrafish larvae requires large-scale data collection and analysis, so our system will play a vital role in such cases. Besides, the system can also be potentially used in other organisms, such as medaka larvae, and other related research, such as drug effects. In the current system, only one single larva was considered in each separate experiment, so our future work will be to achieve the experiments with multiple larvae in large scale for more high-throughput readouts and knowledge discovery.
- 24 August 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 19 (4) , 3007-3019
An Automated Experimentation System for the Touch-Response Quantification of Zebrafish Larvae
Touch-response experimentation in zebrafish helps researchers better understand the link between genetics, drug effects, and behaviors. However, commonly manually conducted experimentation cannot fulfill a high-throughput screening and often delivers low accuracy and lacks reproducibility. Thus, the main aim of this work is to establish a fully automated robot-assisted experimentation system with minimal human participation to conduct the touch-response experimentation with freely swimming zebrafish larvae. Our designed system is able to undertake the role of repeated touch-response experiments at predefined specific location of the larvae in different ages and under different conditions, with high accuracy, robustness, and repeatability, and can also get comparable experimental results. The errors of the detection methods are less than 3 pixels and the offset errors of the touching points are less than 5%. Designed for high-efficiency experimentation, this system will promisingly release a great amount of the burden for the biological operators from touch-response experiments and may also have potential applications in other organisms for touch-evoked response analysis. Note to Practitioners—This article presented an automated touch-response experimentation system on zebrafish larvae, which can release a huge burden of the biological operators by achieving accurate, efficient, and repeated touch-response experiments. Manually conducted experiments are time-consuming and has low accuracy as well as difficult to do the quantification of the behaviors of the larvae. Our proposed system introduced an automated experiment pipeline and can also generate the behavior quantification of the larvae automatically, as the verification experiments confirmed. Basically, the research on the behaviors of zebrafish larvae requires large-scale data collection and analysis, so our system will play a vital role in such cases. Besides, the system can also be potentially used in other organisms, such as medaka larvae, and other related research, such as drug effects. In the current system, only one single larva was considered in each separate experiment, so our future work will be to achieve the experiments with multiple larvae in large scale for more high-throughput readouts and knowledge discovery.