- Speech and Audio Processing
- Music and Audio Processing
- Speech Recognition and Synthesis
- Teleoperation and Haptic Systems
- Tactile and Sensory Interactions
- Stroke Rehabilitation and Recovery
- Gaze Tracking and Assistive Technology
- Engineering Diagnostics and Reliability
- Machine Fault Diagnosis Techniques
- Advanced Data Compression Techniques
- Gear and Bearing Dynamics Analysis
- Logic, Reasoning, and Knowledge
- AI-based Problem Solving and Planning
- Hand Gesture Recognition Systems
- Fault Detection and Control Systems
- Hearing Loss and Rehabilitation
- Machine Learning and Algorithms
- Interactive and Immersive Displays
- Robotic Path Planning Algorithms
- Motor Control and Adaptation
- Digital Filter Design and Implementation
- Wireless Communication Networks Research
- Tribology and Lubrication Engineering
- Mechanical Engineering and Vibrations Research
- Virtual Reality Applications and Impacts
Anhui Polytechnic University
2024
Zhongyuan University of Technology
2014-2022
Ruhr University Bochum
2020-2022
Beijing Institute of Technology
2019-2022
Zhengzhou Institute of Machinery
2022
Chengdu University of Information Technology
2017
Sichuan University
2017
University of South Florida
2002-2006
City University of Hong Kong
2002
In telemanipulation systems, assistance through variable position/velocity mapping or virtual fixture can improve manipulation capability and dexterity [3, 5, 6, 7, 8]. Conventionally, such is based on the sensory data of environment without knowing user's motion intention. this paper, intention combined with real-time information for applying appropriate assistance. If current task following a path, applied. aligning end-effector target, an attractive force field produced. Similarly, if...
In this paper, we propose a tabletop augmented reality (AR) three-dimensional (3D) display system based on integral imaging by using holographic optical element (HOE) that performs the function of microlens array. The array HOE records wavefronts set tilted spherical waves theory reflective volume holograms. reconstruction, an elemental image projected projector and collimated relay optics satisfies Bragg matching conditions reconstructs waves. Based imaging, 3D is generated tilting in...
A novel method to solve the rotating machinery fault diagnosis problem is proposed, which based on principal components analysis (PCA) extract characteristic features and Morlet kernel support vector machine (MSVM) achieve classification. Firstly, gathered vibration signals were decomposed by empirical mode decomposition (EMD) obtain corresponding intrinsic function (IMF). The EMD energy entropy that includes dominant information defined as features. However, extracted remained...
This paper describes the development of intelligent mapping from a haptic user interface to remote manipulator assist individuals with disabilities performing vocational tasks. mapping, referred an assistance function, is determined on basis environmental model or sensory data guide motion telerobotic while given task. Human input enhanced rather than superseded by computer. Three manual dexterity assessment tests, commonly used in occupational therapy field, were chosen implement several...
In order to effectively recognize the bearing’s running state, a new method based on kernel principal component analysis (KPCA) and Morlet wavelet support vector machine (MWSVM) was proposed. First, gathered vibration signals were decomposed by empirical mode decomposition (EMD) obtain corresponding intrinsic function (IMF). The EMD energy entropy that includes dominant fault information is defined as characteristic features. However, extracted features remained high-dimensional, excessive...
For many small- and medium-vocabulary tasks, audio-visual speech recognition can significantly improve the rates compared to audio-only systems. However, there is still an ongoing debate regarding best combination strategy for multi-modal information, which should allow translation of these gains large-vocabulary recognition. While integration at level state-posterior probabilities, using dynamic stream weighting, almost universally helpful small-vocabulary systems, in recognition, accuracy...
End-to-end acoustic speech recognition has quickly gained widespread popularity and shows promising results in many studies. Specifically the joint transformer/CTC model pro-vides very good performance tasks. However, under noisy distorted conditions, still degrades notably. While audio-visual can significantly improve rate of end-to-end models such poor it is not obvious how to best utilize any available information on visual signal quality reliability these models. We thus consider...
With the growing availability of smart devices and cloud services, personal speech assistance systems are increasingly used on a daily basis. Most redirect voice recordings to central server, which uses them for upgrading recognizer model. This leads major privacy concerns, since private data could be misused by server or third parties. Federated learning is decentralized optimization strategy that has been proposed address such concerns. Utilizing this approach, on-device training....
This paper describes the Hidden Markov Model (HMM) based skill learning and its application in a motion therapy system using haptic interface. A relatively complex task, requiring along labyrinth is used. normal subject executes this task for number of times best trajectory selected as learned skill, which considered virtual therapist who can train persons with disabilities to complete task. Two on upper limb (cerebral palsy) were trained therapist. The performance before after training,...
Audio-visual speech recognition (AVSR) can significantly improve performance over audio-only for small or medium vocabularies. However, current AVSR, whether hybrid end-to-end (E2E), still does not appear to make optimal use of this secondary information stream as the is clearly diminished in noisy conditions large-vocabulary systems. We, therefore, propose a new fusion architecture-the decision net (DFN). A broad range time-variant reliability measures are used an auxiliary input...
The temperature of the bevel gear tooth surface will increase greatly because its special structure. Modal is characteristic reaction In order to design and manufacture higher quality aerial gears, it necessary study modal gears under action field. Firstly, a field calculation model established according heat transfer theory. Then, based on traditional linear theory, analysis influence was considering rotational speed stiffness. Based model, driven certain aviation carried out. It can be...
For many small- and medium-vocabulary tasks, audio-visual speech recognition can significantly improve the rates compared to audio-only systems. However, there is still an ongoing debate regarding best combination strategy for multi-modal information, which should allow translation of these gains large-vocabulary recognition. While integration at level state-posterior probabilities, using dynamic stream weighting, almost universally helpful small-vocabulary systems, in recognition, accuracy...
We discuss the possibility of improving eyehand coordination in children diagnosed with this problem, using a robotic mapping from haptic user interface to virtual environment. Our goal is develop an assessment and training procedure that will result handwriting taking advantage force feedback provided by device. Force can be used guide subject's hand predetermined trajectory when he/she unable move response visual feedback. also incorporate inertia viscosity effects decrease tremor as well...
<title>Abstract</title> Class imbalance inevitably occurs in dynamic data stream scenarios and can pose tremendous challenges for mining. To address these challenges, an adaptive resampling weighted ensemble method (ARWE) is proposed this paper. First, the subdivision Poisson (DSPR) module ARWE developed to class problem thedata stream. DSPR combines local information from minority samples with rate design a sample-weighting scheme that enhance visibility of samples, particularly those at...
In telemanipulation systems, assistance of virtual fixture can improve manipulation capability and dexterity. This provides aids not only for path following, but also reaching target avoiding obstacles. Conventionally, these assistances are based on the environment information, without knowing user's motion intention. this paper, intention is combined with real-time information applying appropriate assistance. If current task following a path, hard orthogonal to applied. Or if position...
Audio-visual speech recognition (AVSR) can effectively and significantly improve the rates of small-vocabulary systems, compared to their audio-only counterparts. For large-vocabulary however, there are still many difficulties, such as unsatisfactory video accuracies, that make it hard over baselines. In this paper, we specifically consider scenarios, focusing on task LRS2 database, where performance is far superior video-only making an interesting challenging setup for multi-modal...
Multi-component non-stationary vibration signals produced at local gear fault can easily be covered by periodic harmonic signal and strong background noise, thus causing difficulty in selecting features diagnosing state.To address this issue, a diagnosis method based on variational mode decomposition (VMD) envelope spectrum was proposed study, which then employed to select gear-fault from multi-component adaptively, extract characteristic frequency, determine health conditions of the...
This paper describes the development of intelligent mapping from a haptic user interface to remote manipulator assist individuals with disabilities performing manipulation tasks. mapping, referred an assistance function, is determined on basis environmental model or sensory data guide motion telerobotic while given task. Human input enhanced rather than superseded by computer. Three manual dexterity assessment tests commonly used in occupational therapy field were chosen implement several...
In this paper, we consider the plan recognition problem in real-time strategy game. A probabilistic algorithm is proposed to predict future goals and identify temporal logic tasks of non-cooperative agent based on observations. order model tasks, library composed Finite Transition System Nondeterministic Büchi Automation. Specially, provide a unified framework combine planning, propose probability calculation calculate posterior distribution tasks. Finally, verify effectiveness by compared...
This paper studies the plan recognition problem of multi-agent systems with temporal logic tasks. The high-level tasks are represented as linear (LTL). We present a probabilistic algorithm to predict future goals and identify agent based on observations their states actions. subsequently build library composed Nondeterministic Bu¨chi Automation model also propose Boolean matrix generation map trajectories task parse matrix. Then, probability calculation formula is proposed calculate...
A 2.0 kbps high quality speech coder is presented in the paper. The coding technique based on multiband excitation (MBE) model which efficient modelling of excitation. spectral envelope by linear predictive (LPC) quantization LPC parameters employs a 2-stage split residual vector (SRVQ) scheme. Performance comparisons with other schemes using discrete cosine transform DCT) and two-dimensional differential LSP (2DdLSP) are given.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML"...
End-to-end acoustic speech recognition has quickly gained widespread popularity and shows promising results in many studies. Specifically the joint transformer/CTC model provides very good performance tasks. However, under noisy distorted conditions, still degrades notably. While audio-visual can significantly improve rate of end-to-end models such poor it is not obvious how to best utilize any available information on visual signal quality reliability these models. We thus consider question...