Van Thanh Huynh
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Van Thanh Huynh has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Hieu Trinh
School of Engineering, Deakin University, Waurn Ponds, Geelong, VIC 3216, Australia
12 shared publications - Dinh Cong Huong
Industrial University of Ho Chi Minh City
7 shared publications Hieu Trinh
School of Engineering Deakin University Australia
3 shared publications- Ali Reza Sattarzadeh
School of Engineering, Deakin University, Waurn Ponds, VIC 3216, Australia
2 shared publications - Amanullah Maung Than Oo
School of Engineering, Macquarie University, Sydney, NSW 2109, Australia
2 shared publications Pubudu N. Pathirana
School of Engineering Deakin University Geelong Victoria Australia
2 shared publications
Latest Publications
- Open Access
- 11 April 2026
- Published by Elsevier in SSRN Electronic Journal
Disturbed Delay-Aware Reinforcement Learning
Reinforcement Learning (RL) is now one of the popular control methods used in the area of robotics, which extends to various robotic domains including applications with teleoperation. Considering such applications, it is inevitable that random communication delay exist in between agent-environment interactions. This delay inevitably causes latency in capturing observations and actions. Moreover, noise can occur through external disturbances in RL entities such as rewards which will result in perturbed values. Tackling complex environments in such a scenario can be challenging as the agent receives noisy measurements amid possible delays. We take measures to address this issue in delay-aware RL by proposing a novel reward correcting filter into the delay handling RL framework. The proposed algorithm will estimate the original reward using available perturbed reward values, smoothing out the noise values present in the reward, making an accurate reward estimate available to the agent. We theoretically prove how the proposed filter can estimate the reward with regards to the monotonicity and convergence of the delay-aware RL framework, which in turn will tackle complex environments like robotic tasks. The validation of the algorithm is carried out with experimentations using the DeepMind Control Suite, which highlighted significant performance by showcasing the new algorithm’s robustness with perturbed rewards under delay-aware RL frameworks.
- 11 April 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessDisturbed Delay-Aware Reinforcement Learning
Reinforcement Learning (RL) is now one of the popular control methods used in the area of robotics, which extends to various robotic domains including applications with teleoperation. Considering such applications, it is inevitable that random communication delay exist in between agent-environment interactions. This delay inevitably causes latency in capturing observations and actions. Moreover, noise can occur through external disturbances in RL entities such as rewards which will result in perturbed values. Tackling complex environments in such a scenario can be challenging as the agent receives noisy measurements amid possible delays. We take measures to address this issue in delay-aware RL by proposing a novel reward correcting filter into the delay handling RL framework. The proposed algorithm will estimate the original reward using available perturbed reward values, smoothing out the noise values present in the reward, making an accurate reward estimate available to the agent. We theoretically prove how the proposed filter can estimate the reward with regards to the monotonicity and convergence of the delay-aware RL framework, which in turn will tackle complex environments like robotic tasks. The validation of the algorithm is carried out with experimentations using the DeepMind Control Suite, which highlighted significant performance by showcasing the new algorithm’s robustness with perturbed rewards under delay-aware RL frameworks.
- Open Access
- 2 December 2025
- Published by Elsevier in Robotics and Autonomous Systems
- Vol. 197
Deterministic delay-aware reinforcement learning
- 2 December 2025
- Published by Elsevier in Robotics and Autonomous Systems
- Vol. 197
Open AccessDeterministic delay-aware reinforcement learning
- Open Access
Observer-based Differentiators for Noisy Signals
We present a collection of different types of observation systems that work as differentiators. These observer-based differentiators can produce estimates for derivatives of a given signal, even though the given signal is prone to noise.
Open AccessObserver-based Differentiators for Noisy Signals
We present a collection of different types of observation systems that work as differentiators. These observer-based differentiators can produce estimates for derivatives of a given signal, even though the given signal is prone to noise.
- Open Access
- 13 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 87048-87067
Traffic State Estimation With Spatio-Temporal Autoencoding Transformer (STAT Model)
Traffic state estimation is an essential component of Intelligent Transportation System (ITS) designed for alleviating traffic congestion. As traffic data is composed of intricate information which can also be impacted by various factors, scholars have been attempting to utilize state-of-the-art deep learning forecasting models in recent years. However, a more complex and robust model is required to extract long-range correlations with large-scale traffic data sequences. In order to overcome the weaknesses of deep learning models, the superior performance of transformers is expected to address this effectively in time-series forecasting with transport data. Employing the capabilities of transformers in extracting long-term trends and dynamic dependencies, the proposed model improves the deep learning prediction performance for real datasets. The findings indicate that the transformer-based model exhibited promising performance in forecasting long-term traffic patterns and characteristics with a large quantity of data. In this paper, a comparison across conventional hybrid deep learning models with the Spatio-Temporal Autoencoder Transformer (STAT) model was conducted using real-world datasets. The multi-head attention-based transformer model outperformed all other comparative approaches for large-scale data demonstrating its importance in measuring the error criteria. Comprehensive evaluations across various traffic prediction datasets have established the validity of the proposed approach. Also, for an efficient model selection, Akaike Information Criterion (AIC), Schwarz Bayesian Information Criterion (SBIC), Hannan-Quinn Information Criterion (HQIC), and corrected AIC (AICc) tools were used to evaluate and compare models based on their ability to balance fit and complexity. The proposed model has shown improvements in RMSE for the STREET I980 and PEMSBAY datasets of 1.03 and 0.27, respectively. In RMSE for 60 minutes, for the same datasets, the improvements were 0.84 for STREET I980 and 0.56 for PEMS-BAY. These findings underscore the potential of transformer-based models in enhancing the performance of traffic prediction systems.
- 13 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 87048-87067
Open AccessTraffic State Estimation With Spatio-Temporal Autoencoding Transformer (STAT Model)
Traffic state estimation is an essential component of Intelligent Transportation System (ITS) designed for alleviating traffic congestion. As traffic data is composed of intricate information which can also be impacted by various factors, scholars have been attempting to utilize state-of-the-art deep learning forecasting models in recent years. However, a more complex and robust model is required to extract long-range correlations with large-scale traffic data sequences. In order to overcome the weaknesses of deep learning models, the superior performance of transformers is expected to address this effectively in time-series forecasting with transport data. Employing the capabilities of transformers in extracting long-term trends and dynamic dependencies, the proposed model improves the deep learning prediction performance for real datasets. The findings indicate that the transformer-based model exhibited promising performance in forecasting long-term traffic patterns and characteristics with a large quantity of data. In this paper, a comparison across conventional hybrid deep learning models with the Spatio-Temporal Autoencoder Transformer (STAT) model was conducted using real-world datasets. The multi-head attention-based transformer model outperformed all other comparative approaches for large-scale data demonstrating its importance in measuring the error criteria. Comprehensive evaluations across various traffic prediction datasets have established the validity of the proposed approach. Also, for an efficient model selection, Akaike Information Criterion (AIC), Schwarz Bayesian Information Criterion (SBIC), Hannan-Quinn Information Criterion (HQIC), and corrected AIC (AICc) tools were used to evaluate and compare models based on their ability to balance fit and complexity. The proposed model has shown improvements in RMSE for the STREET I980 and PEMSBAY datasets of 1.03 and 0.27, respectively. In RMSE for 60 minutes, for the same datasets, the improvements were 0.84 for STREET I980 and 0.56 for PEMS-BAY. These findings underscore the potential of transformer-based models in enhancing the performance of traffic prediction systems.
- Open Access
- 22 April 2025
- Published by Elsevier in Sensors and Actuators A: Physical
- Vol. 390
Development of a robust closed loop pressure control system for droplet generation
- 22 April 2025
- Published by Elsevier in Sensors and Actuators A: Physical
- Vol. 390
Open AccessDevelopment of a robust closed loop pressure control system for droplet generation
- Open Access
- 22 April 2024
- Published by MDPI AG in Sustainability
- Vol. 16 (8) , 3483
SustainableTransformation of Undergraduate Engineering Education in China through Sino-Australian Cooperation: A Case Study on Electro-Mechanical System
China’s undergraduate engineering education is facing two significant challenges: passive learning and limited cross-cultural communication. In response, active learning methods such as project-based learning (PBL) and Sino-foreign cooperative education emerge as promising solutions. However, despite their potential, PBL’s application remains constrained, and many Sino-foreign cooperative programs persist with traditional passive learning approaches. How to adeptly integrate the two, namely realizing active learning in Sino-foreign cooperative education, is the primary objective of this study. To this end, a sustainable strategy for cross-cultural transfer and the adaptation of active learning is proposed. Initially, a PBL-based active learning pedagogy is tailored for a course within a Sino-Australian cooperative undergraduate program. This approach integrates hands-on activities, teamwork, and the flipped classroom method. Moreover, to ensure implementation quality, an assessment method reliant on continuous student survey feedback is suggested. Subsequently, through the statistical analysis of students’ survey data as well as final grades, both quantitative and qualitative results affirm the efficacy of the proposed strategy.
- 22 April 2024
- Published by MDPI AG in Sustainability
- Vol. 16 (8) , 3483
Open AccessSustainableTransformation of Undergraduate Engineering Education in China through Sino-Australian Cooperation: A Case Study on Electro-Mechanical System
China’s undergraduate engineering education is facing two significant challenges: passive learning and limited cross-cultural communication. In response, active learning methods such as project-based learning (PBL) and Sino-foreign cooperative education emerge as promising solutions. However, despite their potential, PBL’s application remains constrained, and many Sino-foreign cooperative programs persist with traditional passive learning approaches. How to adeptly integrate the two, namely realizing active learning in Sino-foreign cooperative education, is the primary objective of this study. To this end, a sustainable strategy for cross-cultural transfer and the adaptation of active learning is proposed. Initially, a PBL-based active learning pedagogy is tailored for a course within a Sino-Australian cooperative undergraduate program. This approach integrates hands-on activities, teamwork, and the flipped classroom method. Moreover, to ensure implementation quality, an assessment method reliant on continuous student survey feedback is suggested. Subsequently, through the statistical analysis of students’ survey data as well as final grades, both quantitative and qualitative results affirm the efficacy of the proposed strategy.
- Open Access
- 9 April 2024
- Published by Institution of Engineering and Technology (IET) in IET Control Theory & Applications
- Vol. 18 (8) , 1085-1097
Minimal operation region prediction for networked control robotic manipulators subject to time‐varying delays and disturbances
Due to the disturbances and varying latency, a teleoperated robotic manipulator might not comply with the master control commands. Although prior studies on minimising the impact of network latency and disturbances on teleoperated robots were conducted, there has been very little research on the prediction of minimal operation regions of robotic arms, especially in the worst‐case scenarios when the disturbances and time delays still prevail even after impact minimisation. This study investigates the problem and proposes a novel solution to predicting minimal operation regions of networked control robotic manipulators. The proposed method can be used to forecast safe operation regions in which the manipulators will certainly enter and exclude regions that the robots will never penetrate. Leveraging on a Lyanonov Krasovskii criterion, the method performs region prediction by establishing minimal reachable bounding sets of the nonlinear, perturbed robotic arm's state vectors guided via a time‐varying delay‐dominant network. Though predominantly nonlinear, the entire prediction process is formulated as a tractable Linear Matrix Inequality (LMI) optimisation problem, which can be solved efficiently and effectively. Efficacy of the proposed method is validated with simulations where a simulated robotic arm is distorted with time‐varying delays and disturbances.
- 9 April 2024
- Published by Institution of Engineering and Technology (IET) in IET Control Theory & Applications
- Vol. 18 (8) , 1085-1097
Open AccessMinimal operation region prediction for networked control robotic manipulators subject to time‐varying delays and disturbances
Due to the disturbances and varying latency, a teleoperated robotic manipulator might not comply with the master control commands. Although prior studies on minimising the impact of network latency and disturbances on teleoperated robots were conducted, there has been very little research on the prediction of minimal operation regions of robotic arms, especially in the worst‐case scenarios when the disturbances and time delays still prevail even after impact minimisation. This study investigates the problem and proposes a novel solution to predicting minimal operation regions of networked control robotic manipulators. The proposed method can be used to forecast safe operation regions in which the manipulators will certainly enter and exclude regions that the robots will never penetrate. Leveraging on a Lyanonov Krasovskii criterion, the method performs region prediction by establishing minimal reachable bounding sets of the nonlinear, perturbed robotic arm's state vectors guided via a time‐varying delay‐dominant network. Though predominantly nonlinear, the entire prediction process is formulated as a tractable Linear Matrix Inequality (LMI) optimisation problem, which can be solved efficiently and effectively. Efficacy of the proposed method is validated with simulations where a simulated robotic arm is distorted with time‐varying delays and disturbances.
- 1 April 2024
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part I: Journal of Systems and Control Engineering
- Vol. 238 (7) , 1219-1230
Design of event-triggered natural second-order observers for second-order vector systems
The problem of designing event-triggered natural second-order state observers for second-order vector systems subject to external disturbances is considered in this paper. In this work, state observer is constructed using discrete measurements at instants when an event-triggered condition is satisfied. Due to the errors that arise by using the event-triggered mechanism and the external disturbance, the method in this paper only robustly estimates the state position and the state velocity of the second-order vector observers instead of estimating them asymptotically. Instead of using the method based on a parameter-dependent Lyapunov function in investigating the stability of the dynamic error system of natural second-order observers, a convex optimization problem is established in this paper to give an event-triggered mechanism and minimized levels in evaluating the boundedness of the dynamic error system of the event-triggered natural second-order observers. An automobile suspension and two numerical examples demonstrate the effectiveness of the proposed method.
- 1 April 2024
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part I: Journal of Systems and Control Engineering
- Vol. 238 (7) , 1219-1230
Design of event-triggered natural second-order observers for second-order vector systems
The problem of designing event-triggered natural second-order state observers for second-order vector systems subject to external disturbances is considered in this paper. In this work, state observer is constructed using discrete measurements at instants when an event-triggered condition is satisfied. Due to the errors that arise by using the event-triggered mechanism and the external disturbance, the method in this paper only robustly estimates the state position and the state velocity of the second-order vector observers instead of estimating them asymptotically. Instead of using the method based on a parameter-dependent Lyapunov function in investigating the stability of the dynamic error system of natural second-order observers, a convex optimization problem is established in this paper to give an event-triggered mechanism and minimized levels in evaluating the boundedness of the dynamic error system of the event-triggered natural second-order observers. An automobile suspension and two numerical examples demonstrate the effectiveness of the proposed method.