Authors:Daniel J. L. L. Pinheiro, Jean Faber, Silvestro Micera, Solaiman Shokur Abstract: Human-machine interfaces (HMIs) can be used to decode a user's motor intention to control an external device. People that suffer from motor disabilities, such as spinal cord injury, can benefit from the uses of these interfaces. While many solutions can be found in this direction, there is still room for improvement both from a decoding, hardware, and subject-motor learning perspective. Here we show, in a series of experiments with non-disabled participants, a novel decoding and training paradigm allowing naïve participants to use their auricular muscles (AM) to control two degrees of freedom with a virtual cursor. AMs are particularly interesting because they are vestigial muscles and are often preserved after neurological diseases. Our method relies on the use of surface electromyographic records and the use of contraction levels of both AMs to modulate the velocity and direction of a cursor in a two-dimensional paradigm. We used a locking mechanism to fix the current position of each axis separately to enable the user to stop the cursor at a certain location. A five-session training procedure (20–30 min per session) with a 2D center-out task was performed by five volunteers. All participants increased their success rate (Initial: 52.78 ± 5.56%; Final: 72.22 ± 6.67%; median ± median absolute deviation) and their trajectory performances throughout the training. We implemented a dual task with visual distractors to assess the mental challenge of controlling while executing another task; our results suggest that the participants could perform the task in cognitively demanding conditions (success rate of 66.67 ± 5.56%). Finally, using the Nasa Task Load Index questionnaire, we found that participants reported lower mental demand and effort in the last two sessions. To summarize, all subjects could learn to control the movement of a cursor with two degrees of freedom using their AM, with a low impact on the cognitive load. Our study is a first step in developing AM-based decoders for HMIs for people with motor disabilities, such as spinal cord injury. PubDate: 2023-06-05T00:00:00Z
Authors:Cheng-kun Jia, Yong-chao Liu, Ya-ling Chen Abstract: Face morphing attacks have become increasingly complex, and existing methods exhibit certain limitations in capturing fine-grained texture and detail changes. To overcome these limitation, in this study, a detection method based on high-frequency features and progressive enhancement learning was proposed. Specifically, in this method, first, high-frequency information are extracted from the three color channels of the image to accurately capture the details and texture changes. Next, a progressive enhancement learning framework was designed to fuse high-frequency information with RGB information. This framework includes self-enhancement and interactive-enhancement modules that progressively enhance features to capture subtle morphing traces. Experiments conducted on the standard database and compared with nine classical technologies revealed that the proposed approach achieved excellent performance. PubDate: 2023-06-05T00:00:00Z
Authors:Xiangda Yan, Zhe Zeng, Keyan He, Huajie Hong Abstract: Cooperative autonomous exploration is a challenging task for multi-robot systems, which can cover larger areas in a shorter time or path length. Using multiple mobile robots for cooperative exploration of unknown environments can be more efficient than a single robot, but there are also many difficulties in multi-robot cooperative autonomous exploration. The key to successful multi-robot cooperative autonomous exploration is effective coordination between the robots. This paper designs a multi-robot cooperative autonomous exploration strategy for exploration tasks. Additionally, considering the fact that mobile robots are inevitably subject to failure in harsh conditions, we propose a self-healing cooperative autonomous exploration method that can recover from robot failures. PubDate: 2023-06-05T00:00:00Z
Authors:Kai Zhao, Ruitao Lu, Siyu Wang, Xiaogang Yang, Qingge Li, Jiwei Fan Abstract: A synthetic aperture radar (SAR) image is crucial for ship detection in computer vision. Due to the background clutter, pose variations, and scale changes, it is a challenge to construct a SAR ship detection model with low false-alarm rates and high accuracy. Therefore, this paper proposes a novel SAR ship detection model called ST-YOLOA. First, the Swin Transformer network architecture and coordinate attention (CA) model are embedded in the STCNet backbone network to enhance the feature extraction performance and capture global information. Second, we used the PANet path aggregation network with a residual structure to construct the feature pyramid to increase global feature extraction capability. Next, to cope with the local interference and semantic information loss problems, a novel up/down-sampling method is proposed. Finally, the decoupled detection head is used to achieve the predicted output of the target position and the boundary box to improve convergence speed and detection accuracy. To demonstrate the efficiency of the proposed method, we have constructed three SAR ship detection datasets: a norm test set (NTS), a complex test set (CTS), and a merged test set (MTS). The experimental results show that our ST-YOLOA achieved an accuracy of 97.37%, 75.69%, and 88.50% on the three datasets, respectively, superior to the effects of other state-of-the-art methods. Our ST-YOLOA performs favorably in complex scenarios, and the accuracy is 4.83% higher than YOLOX on the CTS. Moreover, ST-YOLOA achieves real-time detection with a speed of 21.4 FPS. PubDate: 2023-06-02T00:00:00Z
Authors:Dazheng Zhao, Yehao Ma, Jingyan Meng, Yang Hu, Mengqi Hong, Jiaji Zhang, Guokun Zuo, Xiao Lv, Yunfeng Liu, Changcheng Shi Abstract: IntroductionThe time-varying and individual variability of surface electromyographic signals (sEMG) can lead to poorer motor intention detection results from different subjects and longer temporal intervals between training and testing datasets. The consistency of using muscle synergy between the same tasks may be beneficial to improve the detection accuracy over long time ranges. However, the conventional muscle synergy extraction methods, such as non-negative matrix factorization (NMF) and principal component analysis (PCA) have some limitations in the field of motor intention detection, especially in the continuous estimation of upper limb joint angles.MethodsIn this study, we proposed a reliable multivariate curve-resolved-alternating least squares (MCR-ALS) muscle synergy extraction method combined with long-short term memory neural network (LSTM) to estimate continuous elbow joint motion by using the sEMG datasets from different subjects and different days. The pre-processed sEMG signals were then decomposed into muscle synergies by MCR-ALS, NMF and PCA methods, and the decomposed muscle activation matrices were used as sEMG features. The sEMG features and elbow joint angular signals were input to LSTM to establish a neural network model. Finally, the established neural network models were tested by using sEMG dataset from different subjects and different days, and the detection accuracy was measured by correlation coefficient.ResultsThe detection accuracy of elbow joint angle was more than 85% by using the proposed method. This result was significantly higher than the detection accuracies obtained by using NMF and PCA methods. The results showed that the proposed method can improve the accuracy of motor intention detection results from different subjects and different acquisition timepoints.DiscussionThis study successfully improves the robustness of sEMG signals in neural network applications using an innovative muscle synergy extraction method. It contributes to the application of human physiological signals in human-machine interaction. PubDate: 2023-06-02T00:00:00Z
Authors:Qiuyu Cui, Pengfei Liu, Hualong Du, He Wang, Xin Ma Abstract: Mobile robots are widely used in various fields, including cosmic exploration, logistics delivery, and emergency rescue and so on. Path planning of mobile robots is essential for completing their tasks. Therefore, Path planning algorithms capable of finding their best path are needed. To address this challenge, we thus develop improved multi-objective artificial bee colony algorithm (IMOABC), a Bio-inspired algorithm-based approach for path planning. The IMOABC algorithm is based on multi-objective artificial bee colony algorithm (MOABC) with four strategies, including external archive pruning strategy, non-dominated ranking strategy, crowding distance strategy, and search strategy. IMOABC is tested on six standard test functions. Results show that IMOABC algorithm outperforms the other algorithms in solving complex multi-objective optimization problems. We then apply the IMOABC algorithm to path planning in the simulation experiment of mobile robots. IMOABC algorithm consistently outperforms existing algorithms (the MOABC algorithm and the ABC algorithm). IMOABC algorithm should be broadly useful for path planning of mobile robots. PubDate: 2023-06-01T00:00:00Z
Authors:Kaibi Zhang, Guofeng Xu, Ye Kelly Jin, Guanqiu Qi, Xun Yang, Litao Bai Abstract: As a type of biometric recognition, palmprint recognition uses unique discriminative features on the palm of a person to identify his/her identity. It has attracted much attention because of its advantages of contactlessness, stability, and security. Recently, many palmprint recognition methods based on convolutional neural networks (CNN) have been proposed in academia. Convolutional neural networks are limited by the size of the convolutional kernel and lack the ability to extract global information of palmprints. This paper proposes a framework based on the integration of CNN and Transformer-GLGAnet for palmprint recognition, which can take advantage of CNN's local information extraction and Transformer's global modeling capabilities. A gating mechanism and an adaptive feature fusion module are also designed for palmprint feature extraction. The gating mechanism filters features by a feature selection algorithm and the adaptive feature fusion module fuses them with the features extracted by the backbone network. Through extensive experiments on two datasets, the experimental results show that the recognition accuracy is 98.5% for 12,000 palmprints in the Tongji University dataset and 99.5% for 600 palmprints in the Hong Kong Polytechnic University dataset. This demonstrates that the proposed method outperforms existing methods in the correctness of both palmprint recognition tasks. The source codes will be available on https://github.com/Ywatery/GLnet.git. PubDate: 2023-05-26T00:00:00Z
Authors:Kiran K. Karunakaran, Sai D. Pamula, Caitlyn P. Bach, Eliana Legelen, Soha Saleh, Karen J. Nolan Abstract: Acquired brain injury (ABI) is a leading cause of ambulation deficits in the United States every year. ABI (stroke, traumatic brain injury and cerebral palsy) results in ambulation deficits with residual gait and balance deviations persisting even after 1 year. Current research is focused on evaluating the effect of robotic exoskeleton devices (RD) for overground gait and balance training. In order to understand the device effectiveness on neuroplasticity, it is important to understand RD effectiveness in the context of both downstream (functional, biomechanical and physiological) and upstream (cortical) metrics. The review identifies gaps in research areas and suggests recommendations for future research. We carefully delineate between the preliminary studies and randomized clinical trials in the interpretation of existing evidence. We present a comprehensive review of the clinical and pre-clinical research that evaluated therapeutic effects of RDs using various domains, diagnosis and stage of recovery. PubDate: 2023-05-25T00:00:00Z
Authors:Julius Pettersson, Petter Falkman Abstract: Collaborative robots have gained popularity in industries, providing flexibility and increased productivity for complex tasks. However, their ability to interact with humans and adapt to their behavior is still limited. Prediction of human movement intentions is one way to improve the robots adaptation. This paper investigates the performance of using Transformers and MLP-Mixer based neural networks to predict the intended human arm movement direction, based on gaze data obtained in a virtual reality environment, and compares the results to using an LSTM network. The comparison will evaluate the networks based on accuracy on several metrics, time ahead of movement completion, and execution time. It is shown in the paper that there exists several network configurations and architectures that achieve comparable accuracy scores. The best performing Transformers encoder presented in this paper achieved an accuracy of 82.74%, for predictions with high certainty, on continuous data and correctly classifies 80.06% of the movements at least once. The movements are, in 99% of the cases, correctly predicted the first time, before the hand reaches the target and more than 19% ahead of movement completion in 75% of the cases. The results shows that there are multiple ways to utilize neural networks to perform gaze based arm movement intention prediction and it is a promising step toward enabling efficient human-robot collaboration. PubDate: 2023-05-25T00:00:00Z
Authors:Chiara Höhler, Laura Wild, Alexandra de Crignis, Klaus Jahn, Carmen Krewer Abstract: IntroductionVirtual Reality/serious games (SG) and functional electrical stimulation (FES) therapies are used in upper limb stroke rehabilitation. A combination of both approaches seems to be beneficial for therapy success. The feasibility of a combination of SG and contralaterally EMG-triggered FES (SG+FES) was investigated as well as the characteristics of responders to such a therapy.Materials and methodsIn a randomized crossover trial, patients performed two gaming conditions: SG alone and SG+FES. Feasibility of the therapy system was assessed using the Intrinsic Motivation Inventory (IMI), the Nasa Task Load Index, and the System Usability Scale (SUS). Gaming parameters, fatigue level and a technical documentation was implemented for further information.ResultsIn total, 18 patients after stroke (62.1 ± 14.1 years) with a unilateral paresis of the upper limb (MRC ≤4) were analyzed in this study. Both conditions were perceived as feasible. Comparing the IMI scores between conditions, perceived competence was significantly increased (z = −2.88, p = 0.004) and pressure/tension during training (z = −2.13, p = 0.034) was decreased during SG+FES. Furthermore, the task load was rated significantly lower for the SG+FES condition (z = −3.14, p = 0.002), especially the physical demand (z = −3.08, p = 0.002), while the performance was rated better (z = −2.59, p = 0.010). Responses to the SUS and the perceived level of fatigue did not differ between conditions (SUS: z = −0.79, p = 0.431; fatigue: z = 1.57, p = 0.115). For patients with mild to moderate impairments (MRC 3–4) the combined therapy provided no or little gaming benefit. The additional use of contralaterally controlled FES (ccFES), however, enabled severely impaired patients (MRC 0–1) to play the SG.DiscussionThe combination of SG with ccFES is feasible and well-accepted among patients after stroke. It seems that the additional use of ccFES may be more beneficial for severely impaired patients as it enables the execution of the serious game. These findings provide valuable implications for the development of rehabilitation systems by combining different therapeutic interventions to increase patients' benefit and proposes system modifications for home use.Clinical trial registrationhttps://drks.de/search/en, DRKS00025761. PubDate: 2023-05-25T00:00:00Z
Authors:Shiqi Yang, Min Li, Jiale Wang, Zhilei Shi, Bo He, Jun Xie, Guanghua Xu Abstract: IntroductionHemiparesis is a common consequence of stroke that severely impacts the life quality of the patients. Active training is a key factor in achieving optimal neural recovery, but current systems for wrist rehabilitation present challenges in terms of portability, cost, and the potential for muscle fatigue during prolonged use.MethodsTo address these challenges, this paper proposes a low-cost, portable wrist rehabilitation system with a control strategy that combines surface electromyogram (sEMG) and electroencephalogram (EEG) signals to encourage patients to engage in consecutive, spontaneous rehabilitation sessions. In addition, a detection method for muscle fatigue based on the Boruta algorithm and a post-processing layer are proposed, allowing for the switch between sEMG and EEG modes when muscle fatigue occurs.ResultsThis method significantly improves accuracy of fatigue detection from 4.90 to 10.49% for four distinct wrist motions, while the Boruta algorithm selects the most essential features and stabilizes the effects of post-processing. The paper also presents an alternative control mode that employs EEG signals to maintain active control, achieving an accuracy of approximately 80% in detecting motion intention.DiscussionFor the occurrence of muscle fatigue during long term rehabilitation training, the proposed system presents a promising approach to addressing the limitations of existing wrist rehabilitation systems. PubDate: 2023-05-24T00:00:00Z
Authors:Yuanyuan Wang, Yu Gu, Yifei Yin, Yingping Han, He Zhang, Shuang Wang, Chenyu Li, Dou Quan Abstract: Speech emotion recognition is challenging due to the subjectivity and ambiguity of emotion. In recent years, multimodal methods for speech emotion recognition have achieved promising results. However, due to the heterogeneity of data from different modalities, effectively integrating different modal information remains a difficulty and breakthrough point of the research. Moreover, in view of the limitations of feature-level fusion and decision-level fusion methods, capturing fine-grained modal interactions has often been neglected in previous studies. We propose a method named multimodal transformer augmented fusion that uses a hybrid fusion strategy, combing feature-level fusion and model-level fusion methods, to perform fine-grained information interaction within and between modalities. A Model-fusion module composed of three Cross-Transformer Encoders is proposed to generate multimodal emotional representation for modal guidance and information fusion. Specifically, the multimodal features obtained by feature-level fusion and text features are used to enhance speech features. Our proposed method outperforms existing state-of-the-art approaches on the IEMOCAP and MELD dataset. PubDate: 2023-05-22T00:00:00Z
Authors:Federico Sangati, Rui Fukushima Abstract: The Perceptual Crossing Experiment (PCE) has been the object of study for over a decade, and aims at explaining how we perceive, interact with, and understand each other in real-time. In addition to human participant studies, a number of computational models have investigated how virtual agents can solve this task. However, the set of implementation choices that has been explored to date is rather limited, and the large number of variables that can be used make it very difficult to replicate the results. The main objective of this paper is to describe the PCE Simulation Toolkit we have developed and published as an open-source repository on GitHub. We hope that this effort will help make future PCE simulation results reproducible and advance research in the understanding of possible behaviors in this experimental paradigm. At the end of this paper, we present two case studies of evolved agents that demonstrate how parameter choices affect the simulations. PubDate: 2023-05-17T00:00:00Z
Authors:Liangdong Wu, Yurou Chen, Zhengwei Li, Zhiyong Liu Abstract: Intelligent manipulation of robots in an unstructured environment is an important application field of artificial intelligence, which means that robots must have the ability of autonomous cognition and decision-making. A typical example of this type of environment is a cluttered scene where objects are stacked and close together. In clutter, the target(s) may be one or more, and efficiently completing the target(s) grasping task is challenging. In this study, an efficient push-grasping method based on reinforcement learning is proposed for multiple target objects in clutter. The key point of this method is to consider the states of all the targets so that the pushing action can expand the grasping space of all targets as much as possible to achieve the minimum total number of pushing and grasping actions and then improve the efficiency of the whole system. At this point, we adopted the mask fusion of multiple targets, clearly defined the concept of graspable probability, and provided the reward mechanism of multi-target push-grasping. Experiments were conducted in both the simulation and real systems. The experimental results indicated that, compared with other methods, the proposed method performed better for multiple target objects and a single target in clutter. It is worth noting that our policy was only trained under simulation, which was then transferred to the real system without retraining or fine-tuning. PubDate: 2023-05-12T00:00:00Z
Authors:Christian Di Natali, Jesus Ortiz, Darwin G. Caldwell Abstract: Wearable robots are becoming a valuable solution that helps injured, and elderly people regain mobility and improve clinical outcomes by speeding up the rehabilitation process. The XoSoft exosuit identified several benefits, including improvement of assistance, usability, and acceptance with a soft, modular, bio-mimetic, and quasi-passive exoskeleton. This study compares two assistive configurations: (i) a bilateral hip flexion (HA, hips-assistance) and (ii) a bilateral hip flexion combined with ankle plantarflexion (HAA, hips-ankles-assistance) with the main goal of evaluating compensatory actions and synergetic effects generated by the human- exoskeleton interaction. A complete description of this complex interaction scenario with this actuated exosuit is evaluated during a treadmill walking task, using several indices to quantify the human-robot interaction in terms of muscular activation and fatigue, metabolic expenditure, and kinematic motion patterns. Evidence shows that the HAA biomimetic controller is synergetic with the musculature and performs better concerning the other control strategy. The experimentation demonstrated a metabolic expenditure reduction of 8% of Metabolic Equivalent of Task (MET), effective assistance of the muscular activation of 12.5%, a decrease of the muscular fatigue of 0.6% of the mean frequency, and a significant reduction of the compensatory actions, as discussed in this work. Compensatory effects are present in both assistive configurations, but the HAA modality provides a 47% reduction of compensatory effects when considering muscle activation. PubDate: 2023-05-11T00:00:00Z
Authors:Hang Su, Wen Qi, Jiahao Chen, Chenguang Yang, Juan Sandoval, Med Amine Laribi Abstract: Robotics have advanced significantly over the years, and human–robot interaction (HRI) is now playing an important role in delivering the best user experience, cutting down on laborious tasks, and raising public acceptance of robots. New HRI approaches are necessary to promote the evolution of robots, with a more natural and flexible interaction manner clearly the most crucial. As a newly emerging approach to HRI, multimodal HRI is a method for individuals to communicate with a robot using various modalities, including voice, image, text, eye movement, and touch, as well as bio-signals like EEG and ECG. It is a broad field closely related to cognitive science, ergonomics, multimedia technology, and virtual reality, with numerous applications springing up each year. However, little research has been done to summarize the current development and future trend of HRI. To this end, this paper systematically reviews the state of the art of multimodal HRI on its applications by summing up the latest research articles relevant to this field. Moreover, the research development in terms of the input signal and the output signal is also covered in this manuscript. PubDate: 2023-05-11T00:00:00Z
Authors:Changxian Xu, Zhongbo Sun, Chen Wang, Xiujun Wu, Binglin Li, Liming Zhao Abstract: In this article, a tensegrity-based knee mechanism is studied for developing a high-efficiency rehabilitation knee exoskeleton. Moreover, the kinematics and dynamics models of the knee mechanism are explored for bringing about further improvement in controller design. In addition, to estimate the performance of the bionic knee joint, based on the limit function of knee patella, the limit position functionality of the bionic knee joint is developed for enhancing the bionic property. Furthermore, to eliminate the noise item and other disturbances that are constantly generated in the rehabilitation process, a noise-tolerant zeroing neural network (NTZNN) algorithm is utilized to establish the controller. This indicates that the controller shows an anti-noise performance; hence, it is quite unique from other bionic knee mechanism controllers. Eventually, the anti-noise performance and the calculation of the precision of the NTZNN controller are verified through several simulation and contrast results. PubDate: 2023-05-05T00:00:00Z
Open Access journal
[96 journals]
