Search

Time sensitive networks support deterministic schedules over Ethernet networks. Due to their high determinism, high reliability and high bandwidth, they have been considered as a good choice for the backbone network of industrial internet of things. In industrial applications, the backbone network connects multiple industrial field networks together and has to carry massive real-time packets. However, the off-the-shelf time-sensitive network (TSN) switches can deterministically schedule no more than 1024 real-time flows due to the limited number of schedule table entries. The excess real-time flows have to be delivered by best-effort services because the switch only supports the two scheduling services. The best-effort services can reduce average delay, but cannot guarantee the hard real-time constraints of industrial applications. To make the limited number of schedule table entries support more real-time flows, first, we relax scheduling rules to reduce the requirement for schedule table entries and formulate the process of transmitting packets as a satisfiability modulo theories (SMT) specification. Then, we divide the SMT specification into multiple optimization modulo theories (OMT) specifications so that the execution time of solvers can be reduced to an acceptable range. Second, we propose fast heuristic algorithms that combine schedule tables and packet injection control to eliminate scheduling conflicts. Finally, we conduct extensive evaluations. The evaluation results indicate that, compared to existing algorithms, our proposed algorithm requires only one-twentieth the number of schedule entries to schedule the same flow set.

Smart wheelchair plays an important role in rehabilitation training and daily movement of people with high mobility disorders. To ensure that the smart wheelchair can operate stably in an unfamiliar environment, it is necessary to solve the problems of locating, mapping and motion planning. Facing the complex indoor environment of rehabilitation training and the wheelchair model, we propose a new smart wheelchair system design. Based on Simultaneous Localization and Mapping (SLAM) technology, location and mapping can be obtained through the Rao-Blackwellized Particle Filters (RBPF) algorithm. We integrate the sensor data into the proposal distribution to solve the particle degradation problem of RBPF and use selective resampling to improve the algorithm efficiency. According to the wheelchair’s own location and the surrounding environment map, global path planning and local real-time path planning are used to complete the motion planning. Experiments show that the smart wheelchair system we proposed can operate robustly and quickly in a complex indoor environment.

As a part of rehabilitation training, rehabilitation gymnastics plays an important role in restoring the patients’ bodily movement. To facilitate the evaluation of rehabilitation exercises, in this work we propose multi-view on a motion transfer framework which is based on GANs and parameterized human model. Thus we can generate virtual exercise images sampled from the front and back view of one person with the appearance features. Motion transfer methods based on 2D information can hardly deal with the ambiguity of human pose and lack of appearance features. Therefore, this work uses a 3D parametric human model to represent the human body and poses, sampling, and combining pixels from pictures token from both front and back view of a source character, to generate target images under given poses. The whole framework we used can be divided into three stages. 1) In the first stage the 3D human mesh will be reconstructed. 2) In the second stage, transformation maps of multi-view will be calculated, which can show the coordinates of required pixels for sampling and placing. 3) In the third stage, a generated adversarial network is employed to optimize the coarse result from the second stage and make it more realistic. In our experiments, the whole framework achieves good results, especially it is proved that sampling features from both the front and back view can preserve more details of appearance by comparison experiment.

Rehabilitation robots are kinds of robots working together with humans to help patients in functional recovery or to assist people to move normally. Since compliance and safety during a robot-aided therapy period are important, impedance control takes both the effect and procedure in rehabilitation into consideration. Different impedance parameters (for example, inertia, damping ratio, and stiffness) are designed to ensure a robot with desired compliance and safe operation. Flexibility in the selection of those parameters ensures that impedance control makes full use of information collected in robot-aided physical therapy. This paper presents a short review of the implementation of impedance control in lower-limb rehabilitation robots. We first show the basic structure of impedance control. Then, we present an overview of the implementation of various impedance-control methods in lower-limb rehabilitation robots. Finally, we explain the limitations. challenges, and the trend in the field of rehabilitation robotics.

In this paper, we propose a unified real-time framework for gait-based age estimation and gender classification that uses just a single image, which reduces the latency in video capturing compared with the existing methods based on a gait cycle. To cope with the problem of lacking motion information in the input single image, we first reconstruct a gait cycle of a silhouette sequence from the input image via a gait cycle reconstruction network. The reconstructed gait cycle is then fed into a state-of-the-art gait recognition network for feature representation learning, which is further used to obtain the class of the gender and the estimated probability distribution of integer age labels. Unlike the existing methods focusing on the gait sequences captured from the side view, the proposed method is applicable to the gait images from an arbitrary view with a single trained model, which is more suitable for real-world application scenarios (e.g., automatic access control). Stand-alone and client-server online systems were implemented based on the proposed method, which validates the real-time/online property in actual scenes. The experiments on the world’s largest multi-view gait dataset demonstrate the effectiveness of the proposed method, which achieves performance improvement compared with the benchmark algorithms.

In order to improve the high-speed performance of conventional speed prediction control, expand the speed regulation range of permanent magnet synchronous motor (PMSM). A flux-weakening method-based model predictive direct speed control (MP-DSC) is proposed in this paper. In this control strategy, when the motor breaks through the rated speed and runs in the high-speed area, flux-weakening control is achieved to break through the rated speed by controlling the d-axis reference current, which is applied to obtain reference voltage. Based on the principle of deadbeat control, the cost function of conventional speed prediction control including the speed error and current error is converted into a cost function containing only the voltage error. Thus, the weighting factor in conventional MP-DSC method is eliminated in the proposed method, and the flux-weakening control is implemented in the high-speed operation area. Simulation and experiment results demonstrate the feasibility of the proposed method.

In this paper, a novel fractional-order charge-pump-phase-locked-loop (FO-CPPLL) using a fractional-order loop filter (FO-LF) is proposed to improve the overall system performance. The FO-CPPLL is obtained by extending the traditional CPPLL order from integer domain to real domain by using the FO-LF. Specifically, a fractional-order capacitor is applied to replace the integer-order capacitor in the traditional second-order LF, and the constructed FO-LF is inserted between PFD/CP and VCO to form the FO-CPPLL. The FO-LF actualizes fine tuning accuracy in seeking the orders between [0,1] for FO-CPPLL to improve the tracking speed and reducing the system phase loss. Also, the proposed FO-CPPLL can obtain lower jitter and better stability. Finally, simulation results verify the analysis and effectiveness of the proposed FO-CPPLL.

In this paper, we propose a unified convolutional neural network (CNN) framework for robust gait recognition against posture changes (e.g., those induced by walking speed changes). In order to mitigate the posture changes, we first register an input matching pair of gait features with different postures by a deformable registration network, which estimates a deformation field to transform the input pair both into their intermediate posture. The pair of the registered features is then fed into a recognition network. Furthermore, ways of the deformation (i.e., deformation patterns) can differ between the same subject pairs (e.g., only posture deformation) and different subject pairs (e.g., not only posture deformation but also body shape deformation), which implies the deformation pattern can be another cue to distinguish the same subject pairs from the different subject pairs. We therefore introduce another recognition network whose input is the deformation pattern. Finally, the deformable registration network, and the two recognition networks for the registered features and the deformation patterns, constitute the whole framework, named DeformGait, and they are trained in an end-to-end manner by minimizing a loss function which is appropriately designed for each of verification and identification scenario. Experiments on the publicly available dataset containing the largest speed variations demonstrate that the proposed method achieves the state-of-the-art performance in both identification and verification scenarios.

Noising image recognition is an important but challenging task. The images collected by the camera are often accompanied by a variety of noise, which will cause the loss of image information to make recognition difficult. With powerful feature extraction, neural network models can be used to process these complex noising images. However, when the noise model to be learned is complex and the number of learning samples is limited, this may lead to overfitting of the network model. In order to increase the diversity of noising images in limited learning samples and improve the generalization ability of the recognition model, we propose a noise modeling method based on CVAE-GAN. First, the encoder maps a real noising image to hidden layer vectors, which contains noise distribution, image category, and pose information. Then the decoder receives these hidden layer vectors to generate an synthetic image of the specified category and noise level. Finally, during the training process, we combine adversarial loss and smoothl1 loss to make the generated noising image more realistic. Experiments show that compared with other current generative models, our method can more effectively learn the distribution rules of complex noise in real images and generate synthetic images more similar to real noise. These synthetic images generated by our method can can alleviate the overfitting problem caused by insufficient data volume and enhance the generalization ability of image recognition models. By this way, we obtain the state-of-the-art recognition performance on the challenging noising image dataset.

Estimating 6D pose of rigid objects has gained increasing attention as it has become an curcial problem in rapidly growing technology related to robotics, augmented reality and autonomous driving. Therefore, the research on 6D pose estimation technology is of great significance. In this paper, firstly, current position of the field is summarized regarding object pose estimation. We found that deep learning combined with traditional methods can produce better results. Then, pose ambiguity which is an open problem needed further study is raised. Finally, the main problems of the current research and possible development directions are identified.

Aiming at the errors caused by the production and assembly of the robotic arm and the problem of low absolute positioning accuracy, a method of robotic arm calibration and teaching based on binocular vision is proposed in this paper. First, the joints of the robot arm is controled to rotate separately, and the motion trajectory data of the end of the robot arm is recorded by the binocular vision. Then the circle center and circle plane normal vector of each joint axis motion is optimized using Adam algorithm. After identification and calculation, accurate kinematic parameters are obtained, and then kinematically model the manipulator once again. After the calibration is completed, the user can use the teaching rod to define the trajectory under the camera, and the defined trajectory is converted to the robot arm coordinate system for inverse kinematics solution. Finally, the robotic armis controled to reproduce the trajectory. Experimental results show that the proposed method can complete the precise calibration of the kinematic parameters of the robotic arm. After calibration, the average error of the end of the robotic arm is within 1mm, and the absolute positioning accuracy is improved by more than 80%. The method proposed in this paper has low cost and simple operation, and it is easy to be popularized and applied in industrial and desktop robotic arm calibration and teaching.

A 3rd-order Sigma-Delta modulator with a PIλ controller is proposed in this paper to improve the signal to noise ratio (SNR) and system stability. Traditional Sigma-Delta modulator mainly focused on forming an integer high order integrator to improve the SNR, whereas this paper presents a fractional-order controller instead to obtain higher SNR and better stability. Specifically, a 1st-order digital integrator and a fractional-order PIλ controller are inserted between the AFE and the quantizer. The 1st-order digital integrator can increase overall loop order to produce much higher SNR for Sigma-Delta modulator and the fractional-order PIλ controller allows fine tuning accuracy in seeking the coefficients for Sigma-Delta modulator and provides stability in the placement of poles and zeros for compensating the phase delays caused by system integrator parts. The parameters of PIλ controller are optimized by using swarm intelligent optimization algorithm. The performance and stability of the proposed 3rd-order Sigma-Delta modulator with PIλ controller are analyzed and demonstrated, respectively.

In this paper, a new online junction temperature monitoring approach for SiC MOSFETs is proposed by measuring the turn-on delay time using the system microcontroller and intelligent gate drive. Specifically, the intelligent gate drive converts the turn-on delay time to the pulse-width of a digital signal. Then the high-resolution capture module in the microcontroller (with 300 picoseconds resolution) is utilized to precisely measure the turn-on delay time. To improve the measurement sensitivity, the intelligent gate drive circuit implements a large gate resistance during the measurement period and immediately switches back to low gate resistance value for normal operations. A prototype is built, and the proposed circuit is tested in both double pulse switching tests and continuous converter operations. From the experimental results, an accurate junction temperature measurement is achieved in real-time with an error less than 1°

Virtualized datacenters are often designed from scratch with multiple, redundant paths. Yet the majority of the existing congestion control schemes for virtual machines or containers are variants based on single-path TCP design. Lacking the flexibility of leveraging underlying paths, these schemes cannot further improve the utilization of datacenter networks, or mitigate hotspot links. In this paper, we examine the performance of multipath congestion control design on typical hypervisor and container virtualization platforms. We observe that, the involvement of virtual switch, together with the multi-tenancy nature on these platforms, poses new challenges when handling multipath traffic. Through realworld experiments with production-grade applications, we further reveal that, while multipath congestion control increases per-connection throughput and achieves better traffic balancing, it experiences performance degradation when the number of connections is abruptly increased or there exist path-sharing subflows. These issues are due to the enforced QoS policies and interface mapping schemes applied by virtual switch. To this end, we present vMCC, a practical solution which incorporates explicit congestion notification (ECN) support on virtual switches and ECN-aware multipath congestion control algorithms. We show by comprehensive evaluations that vMC-C improves throughput, round trip time, fairness, and energy efficiency for cloud datacenter traffic, and subsequently benefits typical cloud workloads.

Existing gait recognition approaches typically focus on learning identity features that are invariant to covariates (e.g., the carrying status, clothing, walking speed, and viewing angle) and seldom involve learning features from the covariate aspect, which may lead to failure modes when variations due to the covariate overwhelm those due to the identity. We therefore propose a method of gait recognition via disentangled representation learning that considers both identity and covariate features. Specifically, we first encode an input gait template to get the disentangled identity and covariate features, and then decode the features to simultaneously reconstruct the input gait template and the canonical version of the same subject with no covariates in a semi-supervised manner to ensure successful disentanglement. We finally feed the disentangled identity features into a contrastive/triplet loss function for a verification/identification task. Moreover, we find that new gait templates can be synthesized by transferring the covariate feature from one subject to another. Experimental results on three publicly available gait data sets demonstrate the effectiveness of the proposed method compared with other state-of-the-art methods.

Service time is defined as the time taken for a courier to deliver a package to the doorstep of a customer after leaving his vehicle and includes the time taken to return back to his vehicle. Service time is a particularly important parameter in logistics job planning as together with travel time, they determine the number of jobs that can be planned for a day. Traditionally, logistics and transportation companies rely on planners to give a manual estimate of service time for different jobs. However, this process is time consuming and inaccurate. As such, a data-driven, automated Service Time Prediction (STP) method is proposed in this study to provide faster and more accurate service time predictions based on historical data. It does so by first determining historical service time from GPS data by detecting when a service is being conducted. Then, given the historical service times, a KNN regression model is used to provide predictions of service time for new jobs. This study was conducted using two sets of data from 2019 provided to us by two separate logistics and transportation companies based in Singapore. The type of goods delivered are FMCG products.