Publications

A Testing Platform for Flapping-Wing Robots

The validation and optimization of the design of flapping-wing robots (FWRs) are difficult. The wind-tunnel instrument usually used in the performance testing of FWRs is expensive. This work introduces a simple and low-cost testing platform that can be conveniently used to test the FWRs. The platform is mainly composed of a rotating mechanism, a control system, a force and image data acquisition system, and an FWR for testing. The influences of the flapping frequency f, angle of attack α, and incoming flow speed v on FWR performance can be tested with the platform. The force analysis of the platform shows the feasibility of the platform design. A prototype of the platform was built for testing the performance of a bird sized large-scale FWR. The results showed that f, α, and v have evident effects on the lift and thrust of the FWR. The deformation of the wings during flapping could improve the aerodynamic performance of the FWR as the testing results indicated. The platform can be utilized for further optimization of the FWR.

Real-Time Chest Compression Depth Rate, Posture, and Person Detections Using Binocular Vision for CPR Training

Algorithm of Imagined Left-Right Hand Movement Classification Based on Wavelet Transform and AR Parameter Model

Brain-computer interface (BCI) provides new communication and control channels that do not depend on the brain's normal output of peripheral nerves and muscles. In this paper, we report on results of developing a single trial online motor imagery feature extraction method for BCI. The wavelet coefficients and autoregressive parameter model was used to extraction the features from the motor imagery EEG and the linear discriminant analysis based on mahalanobis distance was utilized to classify the pattern of left and right hand movement imagery. The performance was tested by the Graz dataset for BCI competition 2003 and satisfactory results are obtained with an error rate as low as 10.0%.

A novel distributed architecture for building Web-enabled remote robotic laboratories

This paper describes a novel distributed architecture for building Web-enabled remote robotic laboratories. The solution presented here focuses mainly on three aspects: easy and efficient network communication in client-server applications, Internet access of networked sensors, portable architecture based on Java 2 platform. These goals have been achieved by the employment of Java 2 platform and other Web tools. The proposed architecture has been successfully implemented by a demonstration project to run a remote robotic laboratory. Experiment results of the demonstration project prove that the architecture works well and functions to be open, extensible, Web-enabled and platform independent. The original Web site of this demonstration project is offline now. But a video of the demonstration project can be downloaded from http://robot.seu.edu.cn/websensor/.

Jumping aided takeoff: Conceptual design of a bio-inspired jumping-flapping multi-modal locomotion robot

Takeoff is the start of flying. Self-takeoff of flapping-wing robots can improve their practical values in large area surveillance, search and rescue, etc. However, self-takeoff for this kind of robots is still a challenging problem. Inspired by the takeoff of birds and insects assisted by their legs and feet pushing the ground during taking off, this work presents the conceptual design of a jumping-flapping multi-modal locomotion robot to study the feasibility of jumping aided takeoff. A multi-modal robots design method is proposed which makes the robots as light as possible. The jumping, flapping, tail, and driving mechanisms are designed respectively. The sensing and control systems are introduced. The self-takeoff motion planning and aerial maneuvers are studied. The simplified jumping model and the simulation results are also presented. The results of this paper could give guidelines for this kind of robots design.

Internet based gripper teleoperation with random time delay by using haptic feedback and SEMG

Random time delay may cause instability in the internet based teleoperation system. Transparency and intuitiveness are also very important for operator to control the system to accurately perform the desired action, especially for the gripper teleoperation system. This paper presents a new grip force control method of gripper teleoperation system with haptic feedback. The system employs the SEMG signal as the control parameter in order to enhance the intuitive control experience for operator. In order to eliminate the impacts on the system stability caused by random time delay, a non-time based teleoperation method is applied to the control process. Besides, neural network and designed fuzzy logic controller is also utilized to improve this control method. The effectiveness of the proposed method is demonstrated by experiment results.

Readout circuit based on double voltage feedback loops in the two‐dimensional resistive sensor array: design, modelling and simulation evaluation

With one voltage feedback loop, the traditional voltage feedback readout circuit could access all elements in the two‐dimensional (2D) resistive sensor array with the shared row–column fashion but it suffered from the low readout rate problem for in it only one element could be selected and measured at the same time. First, the authors designed a novel readout circuit based on double voltage feedback loops (RC‐DVFLs) which could access two elements in the array simultaneously. Then, an approximate circuit model of the RC‐DVFL was presented and its mathematical equivalent resistance expressions of the elements being tested were analytically derived. Followed, the RC‐DVFL and its expressions were evaluated by simulation experiments and test experiments with a prototype circuit of the RC‐DVFL. The experiment results show that two elements in the 2D resistive sensor array can be accessed with the RC‐DVFL but only part crosstalk is suppressed, and the mathematical equivalent resistance expressions of the elements being tested can be used as the general formulas to evaluate the performance of the RC‐DVFL.

Tactile Elastography

TapeTouch: A Handheld Shape-changing Device for Haptic Display of Soft Objects

Haptic feedback is widely used to enhance realism in virtual reality (VR). Shape and softness are two common factors perceived by the users in the haptic rendering of soft objects. To integrate these factors, we propose a new handheld shape-changing device, TapeTouch, to provide various shapes and softness in real time. TapeTouch is based on a controllable shape-changing tape, which is mainly composed of four motors and a section of brass tape. We design a structure of the components to fit a portable controller and allow to flexibly adjust the shape of the brass tape. After decoding desired shapes into the signals to control the motor, we automatically reproduce varying shapes and levels of softness to the finger or palm touching the shape-changing tape. We conducted two user studies to understand the capability of TapeTouch to render shape and softness, and the results showed that TapeTouch could provide a variety of distinguishable shapes as well as multiple levels of softness. Based on the results, we performed two VR experience studies to verify that the haptic feedback from TapeTouch enhances VR realism.

MH-Pen: A Pen-type Multi-mode Haptic Interface for Touch Screens Interaction

Touch screen technology supplies a new approach to interact with virtual environments. For haptic interaction on a touch screen, haptic devices that are capable of simultaneously conveying tactile and force information to users are highly desired for enhancing the sense of reality and immersion. To this end, a prototype haptic interface, called MH-Pen, was developed and fabricated to display the virtual interactive information through multi-mode haptic feedback. The MH-Pen is a self-contained system that provides vibrotactile feedback and precise force feedback by integrating three types of actuators. In this paper, MH-Pen's design, specifications, and working principle are described. Subsequently, to accurately display the interaction force, a hybrid actuator was designed by combining a piston-type magnetorheological (MR) actuator and a voice coil motor (VCM), and a closed-loop control scheme was built to manage the hybrid actuator. Finally, we objectively and subjectively evaluated the force feedback performance and the effect of multi-mode haptic display of the MH-Pen through physical measurements and psychophysical experiments of virtual surface stiffness display. The results show that improving the precision of force feedback and using multi-mode haptic display are both useful and necessary to enhance the sense of human-computer interaction realism.

Design of a Haptic Interface Based on Cable Robot and Ultrasonic Transducers Array

This paper introduces a haptic interface based on cable robot and ultrasonic transducers array. This novel haptic interface device is capable of invoking compound haptic sensations and displaying multiple properties of virtual objects, including texture, three-dimensional shape, and weight etc. With a network of cables kept in tension, users are able to perceive force feedback in virtual reality applications. 1-DOF grasping and 6-DOF manipulation are provided by the cable-driven parallel mechanism. The kinematic models and static analysis are presented. The tactile feedback is provided by an ultrasonic transducers array. Cable-driven force and ultrasonic tactile feedback are applied to the hand simultaneously. An implementation of the theory and system design is proposed in this paper. Experiments are conducted to confirm the accuracy advantage of our system in the virtual environment. The results of our study indicate that this device has great application potential in human-robot interaction.

Enhancing Robotic Tactile Exploration With Multireceptive Graph Convolutional Networks

While robotic tactile sensors have been developed to help robots to perceive and interact effectively with their surrounding environment by mimicking the structure and function of human skin, most of them overlook the role of near-contact behavior and data structure modeling in robotic perception, which limits robotic exploration capabilities. To address this problem, this article presents a novel proximity-tactile fingertip (PT-TIP) sensor, and a new multireceptive graph convolutional network (MR-GCN) that seamlessly integrates near-contact behavior and tactile perception in rich sensory data. Moreover, MR-GCN utilizes two graph structures, including topology graph and affinity graph, to capture temporal and spatial connections and differences among sensing units on PT-TIP, and it learns a robust feature representation from different receptive fields with attention mechanisms. The performance of MR-GCN was evaluated in two common robotic tasks, namely, object recognition and grasp stability detection, and the results show that the presented method outperforms state-of-the-art work in both tasks.

Velocity field assist-as-needed controller for upper limb rehabilitation with sEMG-based spasticity detection

A Universal Jamming Gripper for Visual-based Grasping in Narrow Spaces

Although grasping objects in narrow spaces is common daily, it is difficult for a robot gripper to fulfill this task. The limited, crowded space and objects of varying shapes and surface properties are major problems, demanding a versatile gripper. This paper uses a lightweight, compact universal jamming gripper to facilitate the grasping operation. A mini camera is mounted on the gripper to capture pictures during manipulation. Besides, the rotation function is integrated into the gripper, allowing it to grasp and rotate simultaneously to remove an object from an environment with barriers and occlusions. The grasping force is tested in experiments to show the process of getting a sphere object. A visual-based grasping approach is presented and evaluated by several grasping experiments. The experiment results show that the gripper can grasp various objects and complete visual-based grasping missions, indicating that the proposed design is adequate and the gripper is versatile.

SleepFC: Feature Pyramid and Cross-Scale Context Learning for Sleep Staging

Automated sleep staging is essential to assess sleep quality and treat sleep disorders, so the issue of electroencephalography (EEG)-based sleep staging has gained extensive research interests. However, the following difficulties exist in this issue: 1) how to effectively learn the intrinsic features of salient waves from single-channel EEG signals; 2) how to learn and capture the useful information of sleep stage transition rules; 3) how to address the class imbalance problem of sleep stages. To handle these problems in sleep staging, we propose a novel method named SleepFC. This method comprises convolutional feature pyramid network (CFPN), cross-scale temporal context learning (CSTCL), and class adaptive fine-tuning loss function (CAFTLF) based classification network. CFPN learns the multi-scale features from salient waves of EEG signals. CSTCL extracts the informative multi-scale transition rules between sleep stages. CAFTLF-based classification network handles the class imbalance problem. Extensive experiments on three public benchmark datasets demonstrate the superiority of SleepFC over the state-of-the-art approaches. Particularly, SleepFC has a significant performance advantage in recognizing the N1 sleep stage, which is challenging to distinguish.

An Efficient Tolerant-Anisotropic Localization for Large-Scale Wireless Sensor Network

Multi-hop localization is a common method which is suitable for large-scale application. However, it is usually influenced by the network anisotropy, leading to the instability of the localization performance. In order to reduce the influence of the network anisotropy on the localization accuracy, this paper regards the localization as a regression forecasting process by constructing the mapping relationship between hop-counts and Euclidean distances among nodes. The method can effectively avoid the influence of anisotropy on localization, and it has low computation overhead and high localization accuracy without setting complex parameters. The simulation experimental results show that compared with the previous similar algorithms, the proposed algorithm can obtain a faster localization speed and a higher localization accuracy.

A New PIS Accelerator for Text Searching

We propose a new design of a hardware accelerator for processing regular expression to speedup text search inside SSD storage (Processing in Storage: PIS). The unique features include parallel processing of 32 streams to quickly identify the first matched character under scan mode and match four characters concurrently under matching mode. In addition, we present a new approach of combining forward and backward scan to accomplish the first character search efficiently. Our experimental results show that the new parallel algorithm reduces the depth of logic circuit and the hybrid architecture performs as well as the Linux Grep algorithm does.

TSA-Former: Linear Transformer with Taylor Series Attention for Sensor-Based Human Activity Recognition

Transformer models have demonstrated superior capability in capturing long-range temporal dependencies crucial for Sensor-Based Human Activity Recognition (HAR). However, the quadratic computational complexity inherent to the Softmax-Attention mechanism significantly impedes their deployment on resource-constrained wearable devices and real-time streaming tasks. To address this, we propose a novel Linear Transformer with Taylor Series Attention specifically tailored for the HAR domain, named TSA-Former. It leverages the first-order Taylor expansion to approximate the Softmax-Attention and utilizes the norm-preserving mapping to approximate the high-order non-linear information, resulting in a linear computational complexity. In addition, TSA-Former integrates a multi-branch architecture featuring multi-scale patch embedding, which enables the model to dynamically capture multi-scale temporal features while minimizing overhead. Experimental results across four public HAR benchmarks, namely UniMiB-SHAR, UCI-HAR, WISDM, and OPPORTUNITY, demonstrate that TSA-Former achieves state-of-the-art (SOTA) accuracy and efficiency, outperforming conventional Transformers and existing linear-attention models. Deployment experiments conducted on the Raspberry Pi 5 platform further validate the model’s superior low-latency and minimal power consumption profile, confirming its robust suitability for real-world embedded HAR applications. Code will be released.