main-content

2022 | Buch

Intelligent Robotics and Applications

15th International Conference, ICIRA 2022, Harbin, China, August 1–3, 2022, Proceedings, Part III

herausgegeben von: Honghai Liu, Zhouping Yin, Prof. Lianqing Liu, Li Jiang, Prof. Guoying Gu, Xinyu Wu, Weihong Ren

Buchreihe: Lecture Notes in Computer Science

Enthalten in:

insite
SUCHEN

Über dieses Buch

The 4-volume set LNAI 13455 - 13458 constitutes the proceedings of the 15th International Conference on Intelligent Robotics and Applications, ICIRA 2022, which took place in Harbin China, during August 2022.

The 284 papers included in these proceedings were carefully reviewed and selected from 442 submissions. They were organized in topical sections as follows: Robotics, Mechatronics, Applications, Robotic Machining, Medical Engineering, Soft and Hybrid Robots, Human-robot Collaboration, Machine Intelligence, and Human Robot Interaction.

Inhaltsverzeichnis

Wearable Sensing and Robot Control

Wearable Ultrasound Interface for Prosthetic Hand Manipulation

Ultrasound can non-invasively detect muscle deformations, which has great potential applications in prosthetic hand control. This research developed a miniaturized ultrasound device that could be integrated into a prosthetic hand socket. This compact system included four A-mode ultrasound transducers, a signal processing module, and a prosthetic hand control module. The size of the ultrasound system was 65 * 75 * 25 mm, weighing only 85 g. For the first time, we integrated the ultrasound system into a prosthetic hand socket to evaluate its performance in practical prosthetic hand control. We designed an experiment to perform six commonly used gestures, and the classification accuracy was $$95.33\%\,\pm \, 7.26\%$$ 95.33 % ± 7.26 % for a participant. These experimental results demonstrated the efficacy of the designed prosthetic system based on the miniaturized A-mode ultrasound device, paving the way for an effective HMI system that could be widely used in prosthetic hand control.

Zongtian Yin, Hanwei Chen, Xingchen Yang, Yifan Liu, Ning Zhang, Jianjun Meng, Honghai Liu
3D Printed Soft Robotic Hand Combining Post-Stroke Rehabilitation and Stiffness Evaluation

Soft rehabilitation devices have been invented and applied for hand function recovery. In this paper, we propose a new Ring-reinforced 3D printed soft robotic hand, which combines hand rehabilitation and joint stiffness evaluation. The elastomer body of Ring-reinforced Soft-Elastic Composite Actuator (R-SECA) is 3D printed directly for fitting different sizes of fingers and the Iterative learning model predictive control (ILMPC) algorithm is used for controlling. Torque compensating layer inside R-SECA enables finger flexion and extension despite finger spasticity. Plastic rings are used to refrain radial expansion and reinforce the actuator. Bending angle and output tip force at different air pressure inputs are explored with four different R-SECA (120 mm, 112 mm, 96 mm, 72 mm length). Four-stroke survivors are recruited to evaluate the effectiveness of the soft robotic hand, and hand function improvement can be observed from the clinical evaluation data and stiffness evaluation outcomes.

Chang Qiu Zhou, Xiang Qian Shi, Zheng Li, Kai Yu Tong
Wearable Sensing Based Virtual Reality Rehabilitation Scheme for Upper Limb Training

Upper limb rehabilitation training is an important method for stroke patients with hemiplegia to restore their upper limb motor ability. The combination of virtual reality (VR) technology and rehabilitation training can increase the effectiveness and interest of the training process. In the process of virtual reality rehabilitation training, dynamic uncertainty factors will affect the effectiveness of rehabilitation training, so it is necessary to adjust the decision of rehabilitation training in real time according to the status of patients. In this paper, a virtual reality upper limb rehabilitation training game with controllable difficulty parameters is designed to collect the electromyographic (EMG) signals and motion signal of patients in the process of movement. Through the fuzzy neural network rehabilitation training decision-making method optimized based on the cuckoo algorithm, the control parameters of virtual reality rehabilitation training scene are adjusted to make the difficulty of rehabilitation training task match the upper limb movement ability of patients adaptively. We recruited 23 stroke patients with different stages of Brunnstrom rehabilitation participated in this experiment. The accuracy of the rehabilitation training decision-making method has an accuracy rate of 96.23%. It can accurately make rehabilitation training decisions, adjust the difficulty of training tasks to a challenging but feasible level, and improve the rehabilitation training effect.

Jialiang Zhang, Yaojie Liu, Juan Liu
Gait Analysis and Phase Recognition Based on Array Fiber Optic Sensing Insole

In order to realize gait assessment and robot-assisted control in lower extremity rehabilitation scenarios, prevention and diagnosis of lower extremity diseases, a sensitivity-enhanced array fiber optic sensing insole for plantar pressure monitoring was designed by taking advantage of the characteristics of fiber optic sensor, such as lightness, anti-electromagnetic interference, strong multiplexing capability, and sensitivity to stress and strain. The gait parameters were effectively analyzed by collecting the plantar pressure data under natural walking. A gait recognition method based on plantar pressure at different walking speeds was proposed to solve the problems of the complexity and poor accuracy of gait recognition. The support vector machine was used to classify four gait periods: the initial double-limb support phase, the single-limb stance phase, the second double-limb support phase and the swing phase. The overall gait phase recognition rate of the classifier was 90.37 $$\%$$ % . The experiment verifies the validity of the fiber optic pressure insole to measure gait parameters and the accuracy of gait recognition.

Nian Peng, Wei Meng, Quan Liu, Shengquan Xie
Gait Phase Detection Based on Time Sequence Adapting to Various Walking Posture and Frequency

This study designs a deep neural network to detect gait phases, including heel-strike (HS), foot-flat (FF), heel-off (HO) and swing (SW). Proposed the concept of “gait image” to be the input and “phase image” to be the output of the model. The model (CFCT) adopts Convolution layers to extract gait-image’s feature, Fully-Connected layers to vary the feature non-linearly and Convolution-Transpose layers to upgrade feature’s dimension to output the phase-image. The CFCT model is capable to predict multi-dimensional gait phases in 1.5 s time sequence and adapt to various walking posture and walking frequency of different people, indicating the model’s robustness. The maximum accuracy of current moment’s gait phase prediction is 98.37%. Results of predicted phases are analyzed according to the time sequence in past 1 s, current moment and future 0.5 s, remaining high and stable accuracy. The maximum accuracy is 96.80% at the time step of future 0.35 s, verifying the effectiveness and stability of the CFCT model.

Siyu Liu, Zhiyong Zhou, Linjun Lu, Xiaohui Xiao, Zhao Guo
In Situ Calibration of a Six-Axis FBG Force/Moment Sensor for Surgical Robot

This paper developed a layered six-axis FBG force/moment sensor to measure the force/moment during robot-assisted bone drilling. The design that eight unique C-shaped beams form a layered elastic structure and eight FBGs are mounted on the structure to sense six-axis force/moment, leads to low chirping risk and low fabrication cost. An in situ calibration method done with a robot and standard weight has been proposed to calibrate the designed sensor and overcome the performance degradation caused by installing and removing sensors. The calibration results that the sensor has excellent measurement accuracy with a relative error of 6.92%. A robot-assisted bone drilling test has been implemented to further prove the potential application of the proposed sensor and the in situ calibration method for high-precision Force/Moment measurement in surgical robots.

Tianliang Li, Fayin Chen, Yifei Su

Wearable Robotics to Characterize, Retrain, and Restore Human Movements

Effects of Brain-Computer Interface and Classical Motor Imagery for Upper Limb Impairment After Stroke: A Case Report

Background. There still exists limitations in the recovery of severe upper limb impairment after stroke, and brain computer interface maybe a hopeful therapy. Methods. A 76-year-old male hemiplegic patient with severe paretic upper limb was admitted. In the first four weeks, 20 sessions classic motor imagery was added in addition to routine treatments. Then, 20 sessions brain-computer interface training was added over the next four weeks. Behavioral characteristics, neuroelectrophysiology and neuroimaging were assessed at multiple time, such as the Fugl­Meyer Assessment Upper Extremity, the Motor Status Scale (MSS), the Action Research Arm Test (ARAT), Active range of motion of the paretic wrist and Modified Barthel Index (MBI). Functional magnetic resonance imaging (fMRI) was used to investigate the effect of the above interventions on the recovery of brain and its structural plasticity. Results. The patient's upper limb motor function improved after two different therapy interventions, however, the efficacy of BCI training was more obvious: after classic motor imagery, the paretic wrist could actively flex, but extension is still irrealizable. However, after BCI training, the paretic wrist was able to extend proactively. The fMRI findings revealed positive and dynamic changes on brain structure and function. Conclusion. BCI training could effectively promote the movement recovery after stroke than traditional motor imagery even if they showed apparent initial paralysis. An association between functional improvement and brain structure remodeling was observed. These findings serve as a conceptual investigations to encourage further relevant research.

Yi-Qian Hu, Rong-Rong Lu, Tian-Hao Gao, Jie Zhuang, Yu-Long Bai
A Synchronous Acquisition System of Ultrasound, sEMG and IMU for Human Motion Prediction

At present, due to the limited information, the single man-machine interface control has some defects in human motion prediction, such as low accuracy and poor robustness. In this work, a multi-modal real-time acquisition system that combines surface electromyography (sEMG), inertial measurement (IMU) and A-mode ultrasound (AUS) information is used to upper limb motion prediction. The device we developed can simultaneously collect three kinds of signals, eliminating the operation of manual alignment. sEMG can reflect the electrical activity of muscle contraction, AUS can detect the deformation of deep muscles, and IMU can obtain information such as the speed and acceleration of the limbs. One healthy subjects participated in the experiment. The results show that the motion prediction accuracy of three modal information fusion is higher than that of any one or two information fusion, which is expected to provide a better control method in exoskeleton or prosthesis.

Yifan Liu, Zongtian Yin, Hongyu Yang, Xingchen Yang, Honghai Liu
Gait Time Parameter Analysis-Based Rehabilitation Evaluation System of Lower Limb Motion Function

Aiming at the problems of low accuracy and high time cost of existing motion function rehabilitation evaluation methods, this paper proposes a rehabilitation evaluation method and system of lower limb motion function based on gait parameter analysis. Firstly, according to the calculation method of gait information related parameters, compare the differences of age and gender in healthy people and the differences of gait information between healthy and disabled group, and preliminarily determine the characteristic parameters of lower limb motion function evaluation. Then, the evaluation indexes are determined through multicollinearity and significance analysis, and the multiple linear regression model of Fugl-Meyer assessment (FMA) score is established. The goodness-of-fit test method is used to prove that the proposed evaluation method can replace the lower limb score of FMA as the evaluation method of lower limb motor function in stroke patients. Finally, the lower limb motor function evaluation system is built in the upper computer, so that the model can be used concretely.

Yue-Peng Zhang, Guang-Zhong Cao, Jiang-Cheng Chen, Ye Yuan, Ling-Long Li, Dong-Po Tan, Zi-Qin Ling
A Portable Fully Coupled Parallel Continuum Manipulator for Nursing Robot: Mechanical Design and Modeling

Robot nurses have become well-known for their contributions to the healthcare industry. Compared with traditional manipulators, cable-driven continuum manipulators have many advantages, such as small manipulator rotational inertia, high dexterity, and large workspace. It is especially suitable for working environment with complex structure and narrow space. In this paper, a portable fully coupled parallel continuum manipulator is designed. The manipulator adopts a spherical parallel mechanism as the basic drive unit. Multiple identical spherical parallel mechanisms are connected in series to form a continuum manipulator. The continuum manipulator installs the motor on the base of the manipulator. To drive the operating platform through the cable, the moment of inertia of the manipulator is reduced, the forward kinematics and inverse kinematics modeling of the basic drive unit are carried out, and the statics modeling of the mechanism is carried out. It solves the problem that the existing traditional manipulator cannot work in a space with a complex structure and a high degree of narrowness, and improves the problems that the existing continuum manipulator has a weak carrying capacity and a small movement space.

Chuanxin Ning, Ting Zhang
A Perturbation Platform and Exoskeleton Simulator for Studying Balance Control of Hip Exoskeleton: Design and Preliminary Validation

In order to deal with the problem of imbalance of the elderly in the process of walking on level ground, this paper proposes a hip exoskeleton with series elastic actuator (SEA) to assist the wearer to adjust the step width, step length and step frequency in real time during walking to maintain balance walking. Through the built multi-sensor fusion gait perturbation and perception integrated platform, the experimenter’s autonomous recovery of balance experiment and exoskeleton-assisted human body's recovery of balance experiment under external wrench state were carried out respectively. The perturbations were applied in four directions to the experimenter while walking, which verified the validity of the balance evaluation model and the assisting effect of the exoskeleton in restoring the balance of human walking. Firstly, according to the established kinematics model, the displacement and velocity of the experimenter's center of mass during walking are calculated, and the system is simplified to a linear inverted pendulum model (LIPM) to evaluate the motion state. Secondly, a balance state evaluation model is established based on the instantaneous capture point (ICP) theory, and the balance state evaluation of the human body is realized by judging the positional relationship between the instantaneous capture point and the support range of the feet. Finally, according to the step strategy in the human body balance strategy, the magnitude of the auxiliary torque when the experimenter is out of balance is calculated through the relative position of the center of mass and the instantaneous capture point. Therefore, it can be shown that the balance evaluation model and exoskeleton control strategy we established can provide the balance restoring torque for the experimenter to swing the leg when the system is about to become unbalanced, which can effectively slow down the unbalanced trend.

Kaixiang Feng, Ting Zhang
Disturbance Observer Compensation Based on Sliding-Mode Approach for Solving Compliant Actuator Tracking Control Problems

The rehabilitation robot needs directly physical interaction in the process of rehabilitation training for patients. Considering the safety of patients, the actuator of the rehabilitation robot should have the advantages of flexibility. Based on the requires of rehabilitation robots, an accurate dynamics model is considered the interaction force between the parts of the compliant actuator, which is established, and the control scheme of the compliant actuator end trajectory tracking is designed. The disturbance observer is designed to estimate the disturbance value and actively eliminate the influence of some disturbances on the compliant actuator. A nonlinear sliding mode controller is designed to reduce the tracking error and shaking the compliant actuator. Through the simulation experiments, compared with the traditional proportional-integral-derivative (PID) controller, it is obvious that the tracking effect of sliding mode control is more efficient on the basis of using the disturbance observer to dispose the disturbance.

Changxian Xu, Jian Gu, Yongbai Liu, Liming Zhao, Zhongbo Sun
Impedance Control of Upper Limb Rehabilitation Robot Based on Series Elastic Actuator

In this paper, to address motor dysfunction caused by factors such as stroke or traffic accidents, a kind of upper limb rehabilitation robot is designed for rehabilitation training. The rehabilitation robot is driven by series elastic actuator (SEA) to make the upper limb rehabilitation robot have flexible output. Flexible output can improve the compliance and safety between the patient and the rehabilitation robot, but impedance control method is needed to ensure the compliance of human–robot interaction. In order to solve the human–robot interaction problem of SEA–based upper limb rehabilitation robot, the dynamic model and an impedance control are established for the SEA–based upper limb rehabilitation robot. The impedance control method of upper limb rehabilitation robot based on terminal position is designed in detail. Aiming at the designed impedance control method, a numerical simulation model is established for the upper limb rehabilitation robot, and the accuracy of the model is verified by the simulation of the upper limb rehabilitation robot. The numerical results show that the impedance controller can meet the needs of the rehabilitation training of the upper limb rehabilitation robot, which improves the coordination of human–robot interaction in the rehabilitation process.

Jian Gu, Changxian Xu, Keping Liu, Liming Zhao, Tianyu He, Zhongbo Sun
Flexible Lightweight Graphene-Based Electrodes and Angle Sensor for Human Motion Detection

Flexible wearable sensors can assist patients with physical injuries or disabilities in auxiliary treatment and rehabilitation, which are of great importance to the development of the future medical field. Most flexible wearable sensors convert physiological signal changes or the changes of body states caused by motion into electrical signals to realize human motion information sensing. Two typical examples are EMG sensors and angle sensors. However, the existing EMG electrodes have many disadvantages such as high manufacturing cost and inferior contact with skin, which makes it impossible to guarantee a stable signal acquisition when the body is in kinetic state. Moreover, angle sensors need to develop in the direction of high sensitivity, ease of use and low cost. Graphene has thus entered the field of vision of researchers. In this work, we tested and analyzed two flexible sensors based on graphene. Firstly, we prepared graphene flexible electrodes and performed human sEMG sensing tests. Meanwhile, we proposed a graphene-based strain gauge with grid structure and performed angle sensing tests. The experimental results show that the graphene electrodes can effectively monitor human movement information such as blink and arm movement with high sensitivity. The graphene grid strain gauge is able to detect flexion angle of joints with high linearity from 20° to 90°. As a flexible sensing material, graphene has the characteristics of high sensitivity, repeatability and ease of use, and can be widely used in different types of sensing, which means that graphene may become one of the prime materials for future wearable sensors.

Wenbin Sun, Quan Liu, Qiang Luo, Qingsong Ai, Wei Meng
The Study of Ankle Assisted Exoskeleton

This article proposes a design of a multidegree-of-freedom ankle assisted exoskeleton based on the gait phase and motion mechanism of human ankle joint. Firstly, the model of the exoskeleton mechanism is designed, and the size parameters of each part are determined through structural analysis. Then, the control strategy of the exoskeleton is analyzed, the gait is obtained by identifying the foot pressure signal through the finite state machine, and the ankle exoskeleton is controlled by matching the corresponding PID parameters according to different gaits. At last, a prototype experiment is carried out. The results show that the ankle exoskeleton can reduce the EMG signal of the calf by 28.91% on average, and has a good effect of rehabilitation.

Yali Han, Jiachen Chang, Zhuangzhuang Jin, Shunyu Liu, Lei Zhou
Motor Learning-Based Real-Time Control for Dexterous Manipulation of Prosthetic Hands

Recent studies on myoelectric-based prosthetic control have shown that surface electromyography (sEMG) can enhance prosthetic intuitiveness by improving motion detection algorithms and continuous data processing. This study aims to use a combination of feature extraction techniques and machine learning approaches to map sEMG signals to 10 upper-limb motions for real-time control. The study implements four machine learning methods (i.e., k-nearest neighbours (k-NN), artificial neural networks (ANN), support vector machines (SVM), linear discriminant analysis (LDA)) as classifiers and six time-domain features (i.e., root mean square (RMS), integrated absolute value (IAV), mean absolute value (MAV), simple square integration (SSI), waveform length (WL), average amplitude change (AAC)) to extract sEMG features to differentiate six individual fingers and four-hand griping patterns. Five subjects volunteered in the research and training datasets were recorded using seven sEMG electrodes for three static and three dynamic arm positions. The modalities were assessed with offline classification performance from the collected datasets and real-time evaluation metrics such as motion completion rate, motion detection accuracies and reach and grasp experiments. Based on the above, the control methodology differentiates independent finger motions with high accuracy, 94% completion rates with 0.23 s data processing and prediction time.

Kemal Balandiz, Lei Ren, Guowu Wei

Robotic Environment Perception

A Method for Object Recognition and Robot Grasping Detection in Multi-object Scenes

Due to robot grasping has always been an open challenge and a difficult problem, it has attracted many researches. With the wide application of deep learning methods in robot grasping, the grasping performance of robots has been greatly improved. Traditionally, many robot grasp detection approaches focus on how to find the best grasp. However, it is very important for robots to have an object recognition function during the grasping process to meet more industrial requirements, such as industrial assembly tasks and sorting tasks. In addition, the problem of missed detection has always existed in the current multi-object grasping detection. To solve the above problems, this paper proposes a two-stage robot grasping method to recognize objects and detect the most likely grasp for every object in multi-object scenes. Our approach achieved a detection accuracy of 65.7% on the VMRD dataset and outperformed the benchmark algorithm by 11.2%. The simulation experimental results showed that our approach achieved a recognition success rate of 96.7% and a grasp success rate of 85% for robotic grasp detection in multi-object scenes.

Jiajun Zheng, Yuanyuan Zou, Jie Xu, Lingshen Fang
Reinforcement Learning for Mobile Robot Obstacle Avoidance with Deep Deterministic Policy Gradient

This paper proposed an improved reinforcement learning (RL) algorithm to develop a strategy for a mobile robot to avoid obstacles with deep deterministic policy gradient (DDPG) in order to solve the problem that the robot spends invalid time exploring obstacles in the initial exploration and speed up the stability and speed of the robot learning. An environment map is used to generate range sensor readings, detect obstacles, and check collisions that the robot may make. The range sensor readings are the observations for the DDPG agent, and the linear and angular velocity controls are the action. The experiment scenario trains a mobile robot to avoid obstacles given range sensor readings that detect obstacles in the map. The objective of the reinforcement learning algorithm is to learn what controls including linear and angular velocity, the robot should use to avoid colliding into obstacles. Simulations results show that the feasibility and certain application value of the method and the algorithm can effectively solve the rewards problem in the process of robot moving, and the execution efficiency of the algorithm is significantly improved. Therefore there are some value of reference and application for development of mobile robot obstacle avoidance system owing to the work of this paper.

Miao Chen, Wenna Li, Shihan Fei, Yufei Wei, Mingyang Tu, Jiangbo Li
An Improved Beetle Antennae Search Optimization Based Particle Filtering Algorithm for SLAM

Particle filter localization technology is the key technology of mobile robot Simultaneous Localization and Mapping (SLAM). To solve the problem of particle degradation in the process of particle pose estimation, an improved optimization strategy of swarm intelligence Beetle Antennae Search (BAS) algorithm is proposed. The key novelty of the proposed particle optimization strategy is that it considers both the randomness and the depth of the particle, improves the global optimization ability of the particle and reduces the RMSE of the particle estimation. The feasibility and effectiveness of the optimization strategy are verified by theoretical comparison and application simulation.

Wei-Dian Ni, Guang-Zhong Cao
Simulation Study of Wireless Coverage in Straight Long Corridors on Container Ship Deck

A wireless coverage simulation based on the ray tracing method is constructed to realize the wireless coverage prediction of the straight long corridor on the container ship deck. The ray reflection model based on the reverse algorithm of the ray tracing method is established to realize the simulation calculation of the point-to-point propagation path. Referring to the recommendations given by the International Telecommunication Union, the Fresnel loss of the straight long corridors and the first Fresnel loss point are calculated, and the ray reflection model is effectively corrected. The least square method is used to fit the double slope of the simulation curve, and the loss formula of the straight long corridors on the deck is obtained. Through experimental simulation, the path loss factors of the fully enclosed corridor are 1.248 and 3.245, and the path loss factors of the semi-closed corridor are 1.251 and 3.444. The simulation results agree with the general conclusions of the ITU recommendation. The simulation can provide the prediction of wireless coverage effect in the design of container ships and provide guidance for the layout design and optimization of base stations of the straight long corridor on container ship deck with different scales.

Yu Zhu, Bing Han
Research on Path Planning Based on the Fusion Algorithm of Adaptive Ant Colony Optimization and Artificial Potential Field Method

It is a hot topic in the field of road landscape planning technology that a mobile robot can quickly and safely find an optimal path in a multi-obstacle environment. In path planning, in light of the problems of poor cooperation and slow convergence of ant colony algorithm in a known environment, the existing potential field method in the local path environment focuses on avoiding dynamic obstacles but cannot guarantee an optimal path. This study provides a new fusion algorithm for path planning optimization in both static and dynamic environments. Firstly, to prevent slipping into a local optimum, create a pheromone diffusion model and adaptively tweak the population information entropy factor to speed up the convergence speed of the Adaptive Ant Colony Optimization (AACO) algorithm. Secondly, on the basis of the globally planned path, by designing the local stability detection and escape functions, the Improved Artificial Potential Field (IAPF) method is utilized to solve the problem of unreachable destination. Finally, we conduct simulation experiments through MATLAB to compare the indicators for evaluating paths, it verifies that the fusion algorithm proposed in this research has obvious advantages in path planning in both static and dynamic environments.

Ran Wang, Qingxin Zhang, Tong Cui, Xinggang Wu
A State-of-the-Art Review on SLAM

SLAM (Simultaneous Localization and Mapping), also known as CML (Concurrent Mapping and Localization), refers to real-time positioning and map building, or concurrent mapping and positioning. After nearly 30 years of research on SLAM, there have been quite a few breakthroughs in the SLAM community. This paper aims to provide an insightful review of information background, recent development, feature, implementation, and recent issue in SLAM. This paper includes the following parts: First of all, it gives an overview of the basic development of SLAM from its introduction to the present. Then, and most importantly, it summarizes the mainstream SLAM technology and theoretical basis. In addition, some cutting-edge and novel SLAM research results are discussed respectively. Finally, this paper summarizes and introduces some practical applications of SLAM technology.

Xuewei Zhou, Ruining Huang

Swarm Robotic Technology and System in Space and Underwater

Synthesis of One DOF Single-Loop Mechanisms with Prismatic Pairs Based on the Atlas Method

This paper proposes a new method to synthesize one DOF single-loop mechanisms (SLMs) with prismatic pairs based on the atlas method. Compared with the traditional synthesis method of SLMs, this method is simple, intuitive, and has a definite physical meaning. Besides, it is a method that is used to synthesize the SLMs with prismatic pairs. It fills the gap in the synthesis of SLM considering prismatic pairs. All constraint types (including overconstraints and non-overconstraints) of SLMs are analyzed comprehensively. The cases containing the lazy pairs are also discussed. The idea of this method is to give a prismatic pair first, and then synthesize other kinematic pairs based on the condition that the motion of this prismatic pair is not constrained. In this paper, a variety of new models are proposed using this method, which verifies the feasibility of this method.

Yang Zhang, Changqing Gao, Hailin Huang, Bing Li
Research on the Hydrodynamic Calculation of Variable Structure Underwater Vehicle Based on CFD

In order to speed up the construction of ocean intelligence, improving the intelligent operation level of Underwater Vehicles is one of the current research directions. This paper takes the Variable Structure Underwater Vehicle (VS-UV) as the research object. Through statics analysis and direct navigation simulation analysis, it is verified that the VS-UV has the ability of large-range low resistance navigation and short-range stable operation. CFD simulation method was used to simulate the 6-DOF motion of the vehicle in the two states before and after the deformation for hydrodynamic calculation. The least square method was used to fit the data to obtain the hydrodynamic coefficient and establish the dynamics model. The comparison between simulation results and real navigation results proves that the method can be used to obtain a reliable model for the VS-UV.

Xiaomeng Liu, Qifeng Zhang, Qiyan Tian, Yaxing Wang, Xuejiao Yang, Dehao Li, Xiaohui Wang
SOINS: A Real-Time Underwater Sonar-Inertial System for Pipeline 3D Reconstruction

In this paper, we propose a real time underwater sonar-inertial system for pipeline reconstruction. In our approach, the sonar data is preprocessed by the likelihood method, from which we calculate the optimal estimation of obstacles position. The cluster center, feature vector and the feature point of sonar point cloud are computed by our cluster-based method, the feature information are utilized to state estimate and 3D reconstruction. During the state estimation step, an IEKF (iterative extended Kalman filter) are used to fuse IMU and sonar data, which are suitable for real-time calculation on the embedded platforms. The proposed method is validated in underwater experiment, the result shows our method have great performance in real-time and accuracy, in which process time can achieve 53 ms and relative translation error can be reduced to 2.5%. The 3D point cloud reconstructed in our method is shown at the end of article.

Wenzhi Gong, Li Xiao, Tian Zeng, Yuchong Li, Zhigang Sun, Zhuo Wang
Subsea Pipeline Inspection Based on Contrast Enhancement Module

Due to the high turbidity of the water and lack of lighting in deep sea, the image of subsea pipeline are blurred and lack of brightness. In the paper an algorithm is proposed to extract centerline of underwater pipeline using image enhancement and pipeline edge detection. The enhancement module based on the color space transformation is given to improve image contrast. Also the threshold segmentation algorithm is put forward to calculate the parameters of Canny operator for edge detection. The centerline of the pipeline is extracted based on the probabilistic Hough transform. Experimental results show that the proposed algorithm is effective and robust.

Ming Zhao, Lin Hong, Zhen-Long Xiao, Xin Wang
Numerical Investigation on Turbulence Models and the Hydrodynamics of a UVMS

Underwater vehicle-manipulator systems (UVMS) have been widely used in many underwater tasks. In this paper, a numerical investigation of the hydrodynamic performances of a small-size UVMS is carried out by using the computational fluid dynamics (CFD) method. Firstly, based on the standard DARPA SUBOFF model, the performance of several common turbulence models is compared, and the SST k- $$\omega$$ ω turbulence model is adopted in our numerical simulations. Then, the three-dimensional model of the UVMS with a complex mechanical structure is simplified to improve the mesh quality and reduce the calculation cost on the premise of ensuring the accuracy of the results. Finally, based on the CFD method, towing test simulations are carried out to obtain the essential hydrodynamic coefficients of the UVMS.

Hang Xu, Lin Hong, Xin Wang, Ming Zhao
Design of Enveloping Underwater Soft Gripper Based on the Bionic Structure

The ocean has been an important research site for scientists in recent years. Many marine creatures with soft bodies such as sea cucumbers are fragile and easily deformed, so it is difficult when grasping these kinds of targets. In this regard, this study developed an underwater soft robot gripper based on a bionic structure. By imitating the envelope structure of the Venus flytrap, the soft robot gripper was designed with a soft finger envelope plate, and the structure was designed by imitating human finger fingerprints. The large and small pressure chambers of soft fingers were designed with the characteristics of different lengths of the segments. By adopting these bionic elements, the grasping ability of the gripper has been greatly improved. The structural parameter optimization design of soft fingers, based on finite element simulation, has been described detailed in the paper. The declination between the simulation results and the actual results is very small, which proves that the accuracy of the optimization method is high. In the simulated underwater environment, some models and living sea cucumbers were grasped for tests. Finally, the experimental results proved that the soft robot gripper can achieve the goal of stable grasping for different objects.

Jiansong Dou, Daohui Zhang, Yanxu Sun, Xin Fu, Xingang Zhao
Research on Formation Obstacle Avoidance Algorithm of Multiple AUVs Based on Interfered Fluid Dynamical System

This paper focuses on planning a two-dimensional (2-D) obstacle avoidance path for the formation of autonomous underwater vehicles (AUVs) in an ocean environment with complicated static obstacles. Inspired by the natural phenomenon of a flowing stream avoiding obstacles, a novel strategy based on an interfered fluid dynamical system (IFDS) is designed. In view of the particular features of the ocean environment, the obstacles are modeled first. Then, by imitating the phenomenon of fluid flow, the IFDS method is used to quickly plan a smooth and safe path for formation AUVs, which conforms to the general characteristic of the phenomenon that running water can avoid rock and arrive at its destination. Finally, formation control is achieved using rigid graph theory. The planned route serves as a known virtual AUV to the leader AUV. The simulation results show that this method has the characteristics of curve continuity and smoothness, enhances obstacle avoidance effects, and has good performance in obstacle avoidance in 2-D path planning.

Wen Pang, Daqi Zhu, Linling Wang
Research on Thermage Robot System Based on Constant Force Control

Nowadays more and more people pay attention to the medical cosmetology industry. This paper presents a study of thermage robot system. In the first, a thermage robot system via hardware design and software design was built. Then, hand-eye calibration, visual recognition and localization, constant force control algorithm were applied to the system. Finally, the experiments based on facial model and human face were carried out, and the feasibility of the proposed thermage robot system was validated effectively.

Fengyi Liu, Chengtao Yue
An Active Obstacle Avoidance Method

Swarm robots often encounter dynamic obstacles when performing tasks, such as moving objects in the scene or other individuals in the robot group. The traditional passive obstacle avoidance method makes the robots take emergency avoidance behaviour when it is about to encounter obstacles. Hoverer this may destroy the group cooperation behaviour, thereby affecting the efficiency of the system. Active obstacle avoidance perceives a dynamic target and predicts the movement of the target and takes the initiative to avoid obstacles, minimizes the impact of obstacle avoidance on the system’s cooperative behaviour. Considering that the defects in the structural design of swarm robots and the avoidance strategy of swarm robots, it is necessary to focus on active obstacle avoidance of swarm robots that is based on the prediction of dynamic targets. An improved obstacle avoidance method is therefore proposed, which enables robots to avoid both static and dynamic obstacles.

Wei Zhu, Yuanzhe Cui, Pengjie Xu, Yichao Shen, Qirong Tang

Medical Robot

Design of Wireless Force Sensing Module of Vascular Interventional Robot

Force information is an indispensable and essential factor in vascular interventional surgery, which significantly affects the accuracy and safety of the procedure. This paper designs a new force sensing module for the vascular interventional robot, which measures the propulsion resistance and twisting torque during catheter and guidewire actuation. The FEA simulation is used to analyze the stress of the elastic body, and to establish the appropriate structure statically. The voltage signal is amplified, filtered and then transmitted wirelessly via Bluetooth. Finally, the force sensing module is subjected to a static calibration experiment, and the acquired signal is filtered using the Kalman filter algorithm. The experimental results have shown that the force sensing module exhibits good linearity and accuracy.

Zhuang Fu, Jianfeng Yao, Zeyu Fu, Chenxin Sui, Jian Fei
Deep Motion Flow Estimation for Monocular Endoscope

For monocular endoscope motion estimation, traditional algorithms often suffer from poor robustness when encountering uninformative or dark frames since they only use prominent image features. In contrast, deep learning methods based on an end-to-end framework have achieved promising performance by estimating the 6-DOF pose directly. However, the existing techniques overly depend on the mass high-precision labelled 6-DOF pose data, which is difficult to obtain in practical scenarios. In this work, we propose a fast yet robust method for monocular endoscope motion estimation named Deep Motion Flow Estimation (DMFE). Specifically, we propose an innovative Key Points Encoder (KPE) supervised by Speeded-up Robust Features (SURF) flow to extract the salient features of endoscopic images. Aiming to ensure real-time capability, we propose a novel 3D motion transfer algorithm to reduce the computational complexity of the essential matrix. Extensive experiments on clinical and virtual colon datasets demonstrate the superiority of our method against the traditional methods, which can provide visual navigation assistance for doctors or robotic endoscopes in real-world scenarios.

Min Tan, Lijuan Feng, Zeyang Xia, Jing Xiong
Constant Force Control Method of Grinding Device

At present, the traditional robot grinding has some shortcomings in output constant force control. As a result, the output force on the grinding object is frequently instable. Improper force can damage the object during grinding and lead huge economic loss. Therefore, how to improve the accuracy of the output force of robot grinding, has become an urgent problem to be solved. In this paper, aim to improve the grinding force control accuracy, a new control framework which is suitable for cylinder driven grinding device is proposed. The control framework is applied to control the cylinder output force of the grinding device, thereby improving the control ability of the high-precision grinding process robot. In the framework, a PID controller with nonlinear differential gain parameters is used, and parameters are optimized by using the Particle Swarm Optimization Algorithm (PSO). The proposed control method, based on the model of the actual cylinder driven grinding device, is verified in MATLAB. The results show that it controls the actual force of the grinding object near the ideal force accurately. The overshoot of the output force on the grinding object is zero and the system stability is very good.

Jia Wen duo, Jiang Zi feng, Dai Yu
Shape Reconstruction Method for Continuum Robot Using FBG Sensors

In this paper, we propose an improved shape reconstruction method based on FBG sensors. Two optical fibers with five uniformly distributed FBG sensors are bonded on a rod, and the curvature of each measuring point can be measured. The relationship between curvature and the rod arc length is obtained by cubic spline interpolation and partial linear interpolation. The rod is differentiated into sufficiently small constant curvature arcs, then the shape of the rod is reconstructed by means of integration. A fourth-order Bezier curve is used as a virtual bending rod to verify the proposed method. Experiments are conducted to reconstruct the shape of actual bending rod. Simulation and experiment results show that the proposed method has higher shape reconstruction accuracy in most cases.

Licheng Hou, Sikyuen Tam, Xingwei Zhao, Bo Tao
Safety Motion Control and End Force Estimation Based on Angle Information in Robotic Flexible Endoscopy

Precise motion control and clear contact force information play an important role in robot-assisted endoscopy. It allows the surgeon to estimate soft tissue stiffness, understand the anatomy, and enables the robot to perform the surgeon’s movement intentions more accurately. However, the ability of surgeon to perceive contact force information and motion status through the flexible endoscope is severely impaired. In this work, we proposed a rotational joint with torque estimation functions based on angle information to achieve the motion control and end force estimation of the flexible endoscope during the robot-assisted endoscopy. Furthermore, the joint consists of two wheels, which are connected with cables and springs. And then, two encoders have been utilized to check the joint positions. In the meantime, a feedforward PID control strategy has been proposed to realize an accurate position control for the designed joint. According to the force estimation study, the perception performance of the proposed joint was characterized with an excellent linearity error (0.53%), a high resolution (7.778 × 10–3 $${\text{N}} \cdot {\text{mm}}$$ N · mm ) and a wide measurement range ( $$- 500{\text{ N}} \cdot {\text{mm}}$$ - 500 N · mm to $$+ 500{\text{ N}} \cdot {\text{mm}}$$ + 500 N · mm ), while the tracking performance demonstrates high sensitivity of the control strategy. The proposed joint has the capability to observe the external interference by estimating the change of the torque. Thus, the proposed method has important application potential in force detection, force feedback and control strategy with enhanced safety during Natural Orifice Transluminal Endoscopic Surgery (NOTES).

Bo Guan, Xingchi Liu, Zhikang Ma, Jianchang Zhao, Yuelin Zou, Jianmin Li
Design and Modeling of a Lightweight Concentric Tube Robot for Nasopharyngeal Surgery

The millimeter diameter of the concentric tube robot enables it to pass through the human nasal cavity for surgery. However, plenty of concentric tube robots adopt a bulky design scheme, forcing the limited space in the operating room to be occupied. In this paper, a lightweight concentric tube robot for nasopharyngeal surgery is proposed. The robot can be mounted on a 6-DOF robot. The length of the concentric tube robot can be adjusted in real-time to perform surgery in different positions. Then, the curvature of the tubes is determined by analyzing the coupling between the tubes. The effect of the stiffness on the curvature of the tubes is analyzed. The forward kinematics model considering the coupling of the concentric tube robot is established. The simulation showed that the stiffness ratio between the tubes is contrary to the changing trend of coupling levels. Finally, the inverse kinematics model of the concentric tube robot is established using the LM algorithm. A simulation is proposed to prove the feasibility of this algorithm. This paper has implications for the motion control of the concentric tube robot.

Gang Zhang, Hangxing Wei, Peng Qi, Honghui Wang, Hao Cheng, Fuxin Du
Research on Puncture Status Perception of Venipuncture Robot Based on Electrical Impedance

With the development of robotics and bioengineering technologies, venipuncture robots are expected to replace manual venipuncture under the supervision of medical staff and free up scarce medical resources. However, current venipuncture robots are still unable to be applied in medical procedures because they still lack in vivo puncture state sensing. A robotic venipuncture decision method based on bio-tissue electrical impedance is proposed in this paper to solve this problem. In this study, the electrical impedance properties of biological tissues are first analyzed, and a robotic venipuncture decision method is proposed based on this property. Then an improved puncture needle that can measure electrical impedance is introduced. Experiments are conducted on a pork model using a self-developed six-degree-of-freedom venipuncture robot. The experimental results show that successful venipuncture has more than a 35% reduction of electrical impedance, and the venipuncture robot based on this method can successfully enter the venous vessels.

Tianbao He, Chuangqiang Guo, Hansong Liu, Li Jiang
An IMU and EMG-Based Simultaneous and Proportional Control Strategy of 3-DOF Wrist and Hand Movements

Multiple degrees-of-freedom (DOFs) simultaneous and proportional control (SPC) is the trend of the electromyography (EMG)-driven human-machine interfaces. Recent studies demonstrated the capability of the musculoskeletal model in SPC. However, the supinator was a deep muscle and the signal-to-noise ratio (SNR) of the supinator surface EMG signal was relatively low. The musculoskeletal model had poor performance when decoding the joint angles of wrist rotation. This study proposed a control strategy intended to address this issue. The proposed decoder utilized an inertial measurement unit (IMU) to record the residual limb movements. The recorded residual limb movements were used to substitute EMG signals from a pair of agonist-antagonist muscles to control the wrist pronation/supination. Meanwhile, the decoder employed EMG signals for control of MCP flexion/extension and wrist flexion/extension. The EMG signals, IMU data and kinematic data were collected simultaneously from an able-bodied subject. To quantify the performance of the decoder, the Pearson’s correction coefficient (r) and the normalized root mean square error (NRMSE) between estimated and measured angles were computed. The results demonstrated that the decoder provided accurate estimations of wrist rotation while the performance of the decoder was not affected by the simultaneous movements with the MCP and wrist flexion/extension.

Zihao Li, Jianmin Li, Lizhi Pan
Application of Feedforward-Cascade Control in an External Pulling Robot for Nerve Restoration

The regeneration process of severed nerves is extremely slow, and the distance and direction of growth are affected by many factors, making it difficult for doctors to determine the time of suture surgery. Existing treatments, which require doctors to manually pull on an external cord to pull the nerve, are highly uncertain. In this paper, we propose an external pulling robot, which connects the fractured nerve to an implanted tendon sheath system (TSS) and automatically pulls the nerve by an external drive module that can be fixed on the body surface. The external driver module is 16 cm by 8 cm by 6 cm. In this paper, a feedforward link is designed to compensate for friction hysteresis in the transmission process of the tendon sheath system. The traction force, speed, and displacement are precisely controlled by the cascade control strategy. The experimental results show that the control system can effectively reduce the errors in the traction process. This paper is of great significance for reducing the cost of nerve restoration and improving the reliability of nerve traction.

Hongrui Fu, Gang Zhang, Han Zeng, Fuxin Du, Rui Song

Intelligent Co-operation in Mobile Robots for Learning, Optimization, Planning, and Control

Crawling Trajectory Generation of Humanoid Robot Based on CPG and Control

There were various gait generation and motion control methods implemented on humanoid robot, while research on humanoid robot crawling and control seems to be litter. In this paper, we establish the CPG model based on Hopf oscillator and get the trajectory of each joint when the robot crawls. Compared with the traditional CPG method, we use the layered and coupled method to establish the CPG expression of robot joint, which reduces the whole amount of computation. In order to correct the problem of the robot’s climbing deviation, we simplified the robot above the waist joint into an inverted pendulum model, and established a position controller to make the robot crawl forward after being disturbed. To increase the robot’s ability to adapt to different environment, we established a compliance control model for the robot’s arm and estimated the force generated when the robot’s arm collided with the ground. Finally, we performed crawling experiments outdoors, the results show that the robot can pass through an environment with a fluctuation height of 3 cm.

Weilong Zuo, Gunyao Gao, Jingwei Cao, Tian Mu, Yuanzhen Bi
A System Integration Method of Product Data Management Based on UG/NX Secondary Development

An integrated development methodology of UG/NX and product data management (PDM) system is carried out based on the UG/NX secondary development technology, aiming at the needs of the PDM system to manage the product design information exported by CAD software tools and multi-user collaborative operation in the product design link, in order to unify the information and file format in the system and simplify user operation. The integrated system based on the Client/Server (C/S) architecture is designed and implemented to manage various product design documents and information. The basic model, software architecture, function flow and realization method of the integrated system are established and analyzed, and some key interfaces of the integrated system are completed. In addition, the architecture and functions of the integrated system are analyzed and evaluated, and the future development work is presented. This development methodology has certain reference application value in the CAD software secondary development, integrated system architecture establishment, functional flow construction as well as related system implementation.

Kai Wang, Pengfei Zeng, Chunjing Shi, Weiping Shao, Yongping Hao
Research on Feature Matching Based on Improved RANSAC Algorithm

Aiming at the problems that the current RANSAC (Random Sample Consensus) algorithm has too large randomness and is susceptible to external point interference, which leads to the reduction of matching accuracy, an improved RANSAC algorithm combining feature matching confidence and grid clustering is proposed. Firstly, rough matching is carried out by the FLANN algorithm, and confidence analysis is carried out on the coarse matching point pairs, then expanding grid clustering around the high confidence point pairs. Multiple local optimal interior points are screened to optimize the global interior points and improve the matching accuracy of feature points. The experimental results show that the improved RANSAC in this paper increases the existence probability of interior points, avoids too many wrong feature matching affecting the model effect of the homography matrix, and improves the accuracy of feature matching.

Xianfeng Wang, Baitong Wang, Zilin Ding, Tong Zhao
Global Optimal Trajectory Planning of Mobile Robot Grinding for High-Speed Railway Body

Reasonable machining trajectory planning could increase the robotic maneuverability and productivity, which is a research hotspot in the field of robotic grinding, especially for large complicated components. To overcome the machining area planning challenges, an optimal robotic machining trajectory planning approach is presented by creating the robot joint configuration model. To begin, a global trajectory planning approach based on the strong surface consistency of a high-speed railway body is proposed to ensure the continuity of robot motion and the optimal configuration. The high-speed railway body is then divided into different areas to ensure robotic accessibility. Finally, the simulation experiment is employed to obtain the appropriate robotic machining trajectory and working attitude, which effectively enhance robotic accessibility and vastly increase processing efficiency and surface quality in the actual robotic grinding of high-speed railway body.

Xiaohu Xu, Songtao Ye, Zeyuan Yang, Sijie Yan, Han Ding
Design of Control Software for a Reconfigurable Industrial Robot Training Platform

For this reconfigurable robot training platform, the platform is divided into 6 modules using fuzzy clustering theory. We use the STM32F407 equipped with the uCOS II system for programming and divide the system into a communication module, a motor module, and a patrol module. The lower computer accepts the instructions of the upper computer, completes the operation of each servo motor through task scheduling, and feeds back the real-time status. Based on QT, the software is built. The upper computer is divided into task module, communication module, and status module. The json class is used to set the configuration file to set the newly created task. The communication task is carried out with other subsystems through the serial port and Ethernet. Use the FUNAUC robot and Eft robot to communicate with the upper computer, accept and analyze the instructions of the upper computer, complete the corresponding training tasks according to the task number and action number, and give feedback on the completion of the robot arm and the task. The above systems cooperate with each other to complete the function of robot training.

Dianyong Yu, Sai Yingnan Bian, Ye Duan
Observer-Based -Consensus Control of Multi-agent Systems Under WTOD Protocol

In this paper, the observer-based $$H_{\infty }$$ H ∞ -consensus control problem is studied for multi-agent systems (MASs) under the weighted try-once-discard (WTOD) communication protocol. The WTOD protocol is implemented in the channel from observer to controller to schedule the data transmissions among the agents. Under such a protocol, for one agent, only one neighboring agent’s information is received at each time instant. First, the state observers are designed based on the measurement outputs to estimate the states of the agents. Then, consensus controllers are designed based on the estimates from the observers. The sufficient condition is obtained such that the $$H_{\infty }$$ H ∞ -consensus performance index is guaranteed. The parameters of the observers and the controllers are then calculated from certain matrix inequalities. Finally, a numerical example is provided to verify the effectiveness and feasibility of the method proposed in this paper.

Jinbo Song, Yafei Jiao, Fan Yang, Chao Xu
Applications of Kalman Filtering in Time Series Prediction

With the development of big data techniques, various data are accumulated and used for time series prediction. As an optimal estimation algorithm, Kalman filtering (KF) is a useful method in realizing time series prediction for linear systems. In this paper, the characteristics of KF and its derivative algorithms (KFDAs) are analyzed and summarized. The existing application results about KFDAs are reviewed respectively in carrying on time series prediction of wind speed and finance. The available comparison results of KFDAs and neural network models are surveyed and discussed on conducting time series prediction, and it is revealed that KFDAs usually outperform neural network.

Xuegui Li, Shuo Feng, Nan Hou, Hanyang Li, Shuai Zhang, Zhen Jian, Qianlong Zi
Non-fragile Consensus Control for MASs with Dynamical Bias

This paper is concerned with the consensus control problem for a class of multi-agent systems with dynamical bias. In system analysis, dynamical bias is taken into account and an augmented state approach is used to deal with the dynamical bias. At the same time, considering the inaccuracy of the controller implementation caused by the complex and changeable environment, a non-fragile controller is proposed to suppress the resulted influence. The main task of this paper is to design an observer-based controller such that the system can achieve the bounded consensus under the influence of the dynamical bias and the controller perturbation. Furthermore, the observer and controller parameters are solved via inequality technique. In the end, a numerical example is given to check on the effectiveness of the proposed control scheme.

Jinnan Zhang, Dongyan Dai, Xuerong Li, Pengyu Wen
ResNet-BiGRU-Attention Based Facial Expression Analysis Model for a Humanoid Robot

In recent years, the robot industry has developed rapidly. With the advance of technology, robots have played an important role on human life in the world. In order to achieve more and more natural and intelligent human-machine interaction, the facial expression recognition model based on ResNet-BiGRU-Attention is proposed in this paper. Firstly, RNN was introduced on the basis of ResNet, the model learns shallow features through ResNet, and learns deep features through RNN. The authors use BiGRU, an enhanced version of RNN, to learn the features of data from two aspects: forward and backward. Secondly, the authors also use the attention mechanism to enhance the weight of the key features in the facial expression. The RAF-DB facial emotion dataset samples were used to train the proposed model. At last, the bipedal humanoid robot NAO was used as an experimental platform. The simulation experiment shows that the model can identify the 7 basic expressions, including happy, angry, nausea, fear, sadness, surprise, and nature. It achieves 86.02% recognition accuracy rate. The proposed model provides a reliable solution for the facial expression recognition in humanoid robot vision.

Yang Lu, Xiaoxiao Wu, Pengfei Liu, Wanting Liu, Xinmeng Zhang, Yixuan Hou
Overtaking Trajectory Planning Based on Model Predictive Control

Autonomous driving technology can greatly increase road safety and reduce accidents, and has become a hot research topic in academia and industry today. However, traditional vehicle motion planning methods often have difficulty in balancing real-time performance with trajectory quality. This paper designs a trajectory planning method based on model predictive control technology, which transforms the motion planning problem into a quadratic planning problem. Compared with previous methods, the proposed method can generate trajectories that meet the requirements of vehicle kinematics and satisfy the requirements of comfort and energy saving while ensuring obstacle avoidance and real-time, and verifies the feasibility of this method in simulation experiments.

Zihan Yuan, Jun Xu
Switching Adaptive Tracking Control for Manipulator with Average Dwell Time

The $$H_{\infty }$$ H ∞ switching adaptive control method with average dwell time is presented in this paper, in order to solve the uncertain problems such as load changes. Due to the change of the load, the parameters of the manipulator change. A switching system is established to simulate the manipulator system whose parameters change. The average dwell time ensures that the switching speed is slow enough that each subsystem can be stabilized, thereby making the entire switching system stable. Simulations show that, under load changes and input disturbances, the manipulator can achieve progressive tracking with the proposed controller.

Hongmei Zhang, Junpeng Shang
An Inspection Planning Method for Steam Generators with Triangular-Distributed Tubes

With the widespread use of nuclear power, steam generators have been widely used, and their periodic inspection is necessary to ensure reliable operation of steam generators. The planning of the steam generator heat transfer tube inspection robot, i.e., the autonomous operation planning of the robot so that it can inspect the heat transfer tube autonomously and without collision with the environment is a difficult research point for nuclear power plant maintenance robots. Based on the triangular distribution of tube plates, an inspection planning method is proposed in this paper, which includes two parts: task planning and path planning. Based on the distribution of the tube plate and the robot motion, the robot inspection methods are classified into three categories and an evaluation index is proposed for selecting the best planning scheme. Based on the planned robot base inspection positions, an improved A* algorithm is proposed in this paper for robot path planning. To verify the reliability of the proposed algorithm, five sets of experiments are conducted in this paper within a steam generator simulation. The experimental results show that the robot can complete the movement and inspection operations on the tube plate according to the planning results, ensuring full coverage inspection for a given heat transfer tube task. Moreover, after comparing the actual running time of the three planning methods, the robot corresponding to the algorithm proposed in this paper has the shortest running time and the highest working efficiency.

Biying Xu, Xuehe Zhang, Yue Ou, Kuan Zhang, Zhenming Xing, Hegao Cai, Jie Zhao, Jizhuang Fan
Memory-Based STOMP for Local Path Planning

Planning and navigation of mobile robots has always been a challenging problem, which has attracted a large number of scholars, especially the research on local path planners. In order to use the past planning experience to guide future path planning, a memory-based stochastic trajectory optimization for motion planning (M-STOMP) is used to solve the local path planning problem. Firstly, the past path planning experience is continuously used to guide the subsequent planning by using memory, which is a method for continuous planning. Then, STOMP algorithm uses Gaussian distribution to generate some smooth paths in the state space, and uses optimized method to update to get a better path. Finally, this method was tested in four different scenarios which validate the proposed method. This paper gave a method for local path planning from a new perspective.

Wenjie Li, Tao Cao, Yunfan Wang, Xian Guo
Research and Verification of Robot Master-Slave Control Algorithm for Nuclear Power Maintenance Scenarios

In this paper, the master-slave control algorithm is investigated and an experimental verification platform is built to validate the master-slave algorithm and the remote maintenance operation flow for the needs of nuclear power remote maintenance scenarios. The paper adopts an incremental master-slave control algorithm based on position-orientation separation method, which overcomes the isomerism and workspace inconsistency problem between the master and slave while ensuring the same motion trend between the master and slave; Secondly, a variety of position mapping scales are designed to meet the needs of nuclear power maintenance operation tasks, taking into account the efficiency and accuracy of the operation. Meanwhile, a safety assurance mechanism is introduced in the master-slave control architecture to eliminate the situation that the slave robot arm moves violently in a short period of time due to operator’s misoperation. Finally, the effectiveness of the master-slave control algorithm is verified by simulating a nuclear remote maintenance task on the experimental validation platform.

Feng Yang, Haihua Huang, Yanzheng Chen, Weiming Li, Quanbin Lai, Rui Ma, Binxuan Sun, Xingguang Duan
Research on Force Perception of Robot End-Effector Based on Dynamics Model

With the wide application of robots in the industrial field, higher requirements are put forward for the use of robots. The force information at the end of the robot is an important execution information of the robot, and the accuracy of its estimation accuracy directly affects the execution precision of the robot. Aiming at this problem, an accurate dynamic model of six-Dofs robot was established, and the torque changes under different working conditions in the process were analyzed. The dynamics simulation model of the robot was built by the co-simulation of MATLAB and Adams software. According to the collected torque information, it will be transformed into the end contact force information. The accuracy of estimating the end force can reach 98.6%.

Zhongshuai Yao, Ming Hu, Yufeng Guo, Jianguo Wu, Jing Yang

Tactile Robotics

Perceptual Properties of Fingertips Under Electrotactile Stimulation

To sense and represent electrotactile perceptual properties is a crucial milestone in order to achieve intuitive haptics. However, electrotactile perceptual properties are very poorly studied. This study presented an experimental study on the electrotactile perceptual properties of fingertips. A series of experimental paradigms were designed based on self-designed hardware and psychophysical evaluation methods. The detection threshold (DT), pain threshold (PT), just-noticed difference (JND), intensity-quality characteristics and multi-level discrimination ability for pulse amplitude (PA), pulse width (PW) and pulse frequency (PF) have been explored. The experiments verified the individual differences in DT and PT and found that the fingertips were more sensitive to PA and thus more valuable for information encoding. In discrete coding, the recognition accuracy decreases with increasing number of levels, preferably less than 4. The results are expected to provide valuable suggestions for the parameter coding of electrotactile information presentation.

Ziliang Zhou, Yicheng Yang, Jia Zeng, Xiaoxin Wang, Jinbiao Liu, Honghai Liu
A Brief Review Focused on Tactile Sensing for Stable Robot Grasping Manipulation

In this paper, we briefly investigate recently published literature on robot grasping with tactile information to understand the effect introduced by tactile modality and summarize the current issues of tactile sensing. Moreover, this paper consists of a review of slip detection during grasping, a review of grasp stability assessment to estimate the current contact state and a review of regrasp to select appropriate grasp adjustment action. Finally, we discuss the current limitations and deficiencies that prevent researchers from using tactile sensors, making it challenging to incorporate tactile modalities into robot perception and properly utilize tactile information to achieve effective and stable grasp performances. We consider that the pipeline consisting of grasp outcome prediction and grasp action adjustment based on machine learning is an appropriate scheme to make full use of tactile information and its potential in robot grasping tasks. More studies in this field are expected in the future.

Zhenning Zhou, Zhuangzhuang Zhang, Kaiyi Xie, Xiaoxiao Zhu, Qixin Cao
A Soft Neuromorphic Approach for Contact Spatial Shape Sensing Based on Vision-Based Tactile Sensor

Robots with tactile sensors can distinguish the tactile property of the object, such as the spatial shape, in many robotic applications. The neuromorphic approach offers a new solution for information processing to encode tactile signals. Vision-based tactile sensing has gradually attracted attention in recent years. Although some work has been done on proving the capacity of tactile sensors, the soft neuromorphic method inspired by neuroscience for spatial shape sensing is remarkably rare. This paper presented a soft neuromorphic method for contact spatial shape sensing using a vision-based tactile sensor. The outputs from the sensor were fed into the Izhikevich neuron model to emit the spike trains for emulating the firing behavior of mechanoreceptors. 9 spatial shapes were evaluated with an active touch protocol. The neuromorphic spike trains were decoded for discriminating spatial shapes based on k-nearest neighbors (KNN). Three spike features were used: average firing rate (FR), the coefficient of variation of the interspike interval (ISI CV), and the first spike firing time (FST). The results demonstrated the ability to classify different shapes with an accuracy as high as 93.519%. Furthermore, we found that FST significantly improved spatial shape classification decoding performance. This work was a preliminary study to apply the neuromorphic way to convey the tactile information obtained from the vision-based tactile sensor. It paved the way for using the neuromorphic vision-based tactile sensor in neurorobotic applications.

Xiaoxin Wang, Yicheng Yang, Ziliang Zhou, Guiyao Xiang, Honghai Liu
Contact Information Prediction Based on Multi-force Training for Tactile Sensor Array with Elastomer Cover

Tactile perception is essential for the grasping and manipulation of the robotic hand. The contact force and position can be detected by the tactile sensor. Tactile sensors with continuous surface cover can achieve higher detection accuracy in some cases than independent sensing units, but the calibration is also more troublesome. In this paper, we propose a contact information prediction method based on multi-force training using a BP neural network. The contact force could be predicted from the output of the sensor units through the mixed training of force and position, where the corresponding calibration data set is generated by collecting multiple forces at each selected point on the surface of the tactile sensor. The feasibility of the proposed calibration algorithm could be verified by comparison experiments.

Qiang Diao, Wenrui Chen, Yaonan Wang, Qihui Jiang, Zhiyong Li
Hand Rehabilitation Modes Combining Exoskeleton-Assisted Training with Tactile Feedback for Hemiplegia Patients: A Preliminary Study

Within the field of rehabilitation for people with hemiplegia, this paper presents a novel training system for hand function rehabilitation. This training system mainly includes a hand rehabilitation exoskeleton, tactile feedback devices and a virtual reality scene. Tactile feedback devices are designed as electric stimulation slip feedback actuator and pneumatic contact force feedback actuator respectively. The virtual reality scene is a human-computer interaction interface built by Unity 3D. Three rehabilitation training modes including a contact force enhanced rehabilitation mode, a mirror therapy mode and an active rehabilitation mode are proposed to provide different feedback stimulation for patients at different rehabilitation stages to obtain better rehabilitation training effect. Verification experiments were conducted to preliminarily show the feasibility of those modes.

Bo He, Min Li, Guoying He

Co-manipulation System with Enhanced Human-Machine Perception

Behavior Tree Based Dynamic Task Planning Method for Robotic Live-Line Maintenance

The biggest challenge for robotic automation in the field is unexpected events caused by uncertainties in an unstructured environment. In classical task planning, the planning and execution of tasks are separated, which makes it difficult to deal with unexpected events in the execution process. However, robots must be able to handle unexpected events to ensure the smooth execution of tasks in complex dynamic environments. This paper introduces a method for dynamic task planning using behavior trees, which is used for robots to perform live-line maintenance operations on overhead lines in distribution networks. This method realizes the dynamic planning of tasks and the handling of unexpected events by structuring the task behavior tree. Experiments and field operations show that the method can realize complex field operations and have unexpected handling capabilities.

Feng Jiabo, Shi Lirong, Liu Lei, Zhang Weijun
Design of Multi-unit Passive Exoskeleton for Running

The passive exoskeleton used to assist the human gait and reduce the energy consumption of the human body. In this paper, through the mathematical expression of the man-machine mechanical power model and the installation design area, combined with the optimization algorithm, some passive exoskeleton structure designs that can reduce the mechanical power consumption of the whole lower limb of the human body are found. The designed passive exoskeleton enables an average reduction of 4.4% compared to disability metabolism by building experimental prototypes for metabolic experiments.

Nianfeng Wang, Fan Yue, Jiegang Huang, Xianmin Zhang
Force Tracking Impedance Control Based on Contour Following Algorithm

The original impedance control is a main force control scheme widely used in robotic force tracking. However, it is difficult to achieve a good force tracking performance in uncertain environment. This paper introduces a modification of the impedance control scheme which has the adaptability to track the desired force in uncertain environment (in terms of the varying location of the environment relative to the manipulators). The relation function of contact force in adjacent control period is derived to estimate the trajectory inclination angle deviation (IAD) of the manipulator. After that, the contour following algorithm which is implemented under a PID controller is proposed. To achieve force tracking impedance control under uncertain environment location, the movement of the manipulator in one control period is determined by estimating the new velocity vector and calculating the impedance correction online, which is based on the position-based impedance control (PBIC). Experiments was presented for testing the performance of IAD estimation and force tracking.

Nianfeng Wang, Jianbin Zhou, Kaifan Zhong, Xianmin Zhang, Wei Chen
Guidance Method for UAV to Occupy Attack Position at Close Range

Unmanned aerial vehicles (UAVs) have become an important role in modern air combat. Aiming at the requirement that UAVs can autonomously maneuver according to the battlefield information to obtain the advantage of attack position in close-range air combat scenarios, a solution based on the constrained gradient method to solve the optimization guidance index of UAV attack position occupation to control UAV maneuvering is proposed. The UAV kinematic model is used as an algorithm control carrier. According to the actual air combat confrontation situation, it judges whether the distance and angle meet the attack or avoid conditions, and designs corresponding optimization indicators. Based on the input target state information, the constrained gradient method is used to optimize the attack or avoidance indicators. Perform the solution to obtain the guidance instructions, and input the instructions into the corresponding UAV kinematics model to complete the UAV's attack position occupation guidance. According to the engineering application requirements, a typical 1V1 air combat simulation test scenario is established. The simulation results show that the method in this paper can guide the UAV to obtain the advantage of attacking position.

Xu Bingbing, Meng Guanglei, Wang Yingnan, Zhao Runnan
The Variation Characteristic of EMG Signal Under Varying Active Torque: A Preliminary Study

Surface Electromyography (sEMG) or EMG contains a large amount of information about human kinematics and kinetics, and has been applied in different working environments. Devices like exoskeletons, smart bracelet performs better with information from EMG introduced into the system. For example, some rehabilitation exoskeletons designed for subjects suffered from nerve injuries are controlled under the strategy called “assist-as-needed”. In these studies, various methods, especially machine learning, have been used to establish a large number of nonlinear relationships between EMG and kinematics, as well as kinetics. However, some conditions that have not been studied before but occur in the system will lead to errors in the overall response of the control system. In this paper, human muscle tissue is regarded as a device with input and output responses, the relationship between the least squares slope of AEMG (Averaged EMG) and the current change in muscle contraction torque $$\Delta T$$ Δ T is studied when the torque generated by muscle contraction is $$T$$ T , the joint angle is $$\theta$$ θ , and the joint movement angular velocity is $$\omega$$ ω . The established relationship provides a potential closed-loop EMG control pathway from human to machine for human-machine interaction devices.

Boxuan Zheng, Xiaorong Guan, Zhong Li, Shuaijie Zhao, Zheng Wang, Hengfei Li
A VPRNN Model with Fixed-Time Convergence for Time-Varying Nonlinear Equation

Robots are widely used in various engineering fields, and the solution to their trajectory tracking problem has attracted increasing attention. Such a problem can be typically transformed into a time-varying nonlinear equation (TVNE). For complex and high-precision robot trajectory tracking problems, a fast and low-error tracking solution is necessary. Therefore, a varying-parameter recurrent neural network (VPRNN) model with a modified power-type time-varying parameter is proposed for solving TVNE. An improved sign-bi-power function is selected for the activation function, then the VPRNN model achieves fixed-time convergence. Numerical comparisons with the general fixed-parameter recurrent neural network model are performed, which demonstrates the superiority of our VPRNN model. Besides, the proposed VPRNN model is successfully used to solve the trajectory tracking problem of a three-link robot, which shows its feasibility in practical applications.

Miaomiao Zhang, Edmond Q. Wu
Surface Electromyography-Based Assessment of Muscle Contribution in Squatting Movements

The surface EMG signal is an electrophysiological signal generated by muscle contraction and collected by placing electrodes on the skin surface, which contains rich information about muscle function and state. The current exoskeleton design also urgently needs to evaluate the muscle contribution to assist joint or part movement, and this result will directly affect the theoretical design of the exoskeleton. From this perspective, in this paper, we measured the surface EMG data during squatting for 10 test subjects, and set up a control group and an test group, with each control group doing 10 continuous squats and each test group doing 10 continuous squats with a hand weight of 10 kg. In this paper, 60 sets of surface EMG data of squatting movements of 10 test subjects were analyzed, and the contribution of 12 different parts of muscles was evaluated based on covariance matrix, and it was obtained that rectus femoris contributed more than 15% of the 12 muscles without weight-bearing, but the contribution of rectus femoris decreased by 50% under weight-bearing condition, and the contribution of medial femoris and biceps femoris increased significantly to more than 15% of the overall.

Zheng Wang, Xiaorong Guan, Zhong Li, Boxuan Zheng, Hengfei Li, Yu Bai
Energy Gradient Descent Method for Actuation of a Direct-Drive Spherical Robotic Wrist

This paper presents an energy gradient descent method for actuating a spherical robotic wrist which is capable of providing three-DOF rotations in one joint. By formulating the relationship between the magnetic energy for driving the spherical rotor and the current inputs supplied to the motor, an energy gradient descent method is proposed by adjusting the supplied current inputs for shaping a minimum magnetic energy point at the desired rotor state. As a result, the rotor will approach to the desired state automatically without the need of any feedback control laws. The solutions to the supplied currents for shaping a desired energy profile can be computed in real-time with a magnetic-flux-based model. The proposed method has been validated with both numerical simulations and experimental tests performed on a cooperative robotic manipulator equipped with a spherical wrist actuator.

Mengke Li, Yaqing Deng, Kun Bai

Compliant Mechanisms and Robotic Applications

Design and Modeling of a Novel Compliant Ankle Mechanism with Flexible Slider-Crank Limbs

This paper presents the conceptual design and modeling of a novel compliant ankle mechanism, which has flexible slider-crank limbs. Two elastic beams are utilized as the springy elements to connect the sliders and crank, which provides the ankle joint with passive rotational stiffness when two sliders are driven independently. Both the forward and inverse kinetostatic model are derived to determine the equilibrium configuration and the corresponding actuation variables. Besides, the rotational stiffness of the studied ankle mechanism is modeled based on results from the kinetostatic model. Results of stiffness analysis reveal that the proposed ankle joint is capable of varying its rotational stiffness if the sliders are controlled properly. The kinetostatic and stiffness models developed in this paper lay a foundation for stiffness design and prototype development in the future work.

Shujie Tang, Genliang Chen, Wei Yan, Hao Wang

The present work proposes an Adaptive Compliant Control scheme based on a closed-form output-redefined and perturbed dynamic model of a Single-link Flexible Manipulator (SLFM) in Unknown Environment. The control scheme is composed of inner and outer controllers. The inner control is designed based on Two-Time Scale Adaptive Robust Control (TTARC) to ensure fast and precise motion control, while the outer control is based on the impedance dynamics aiming to offer a desired compliant behavior in constrained motion. External force is estimated based on the extended Kalman Filter (EKF). The stability of the closed-loop system is verified through Lyapunov theory. The effectiveness of the overall control scheme is verified through simulation.

Cianyi Yannick, Xiaocong Zhu, Jian Cao
A Novel Discrete Variable Stiffness Gripper Based on the Fin Ray Effect

Variable stiffness grippers can adapt to objects with different shapes and gripping forces. This paper presents a novel variable stiffness gripper (VSG) based on the Fin Ray effect that can adjust stiffness discretely. The main structure of the gripper includes the compliant frame, rotatable ribs, and the position limit components attached to the compliant frame. The stiffness of the gripper can be adjusted by rotating the specific ribs in the frame. There are four configurations for the gripper that were developed in this research: a) all ribs OFF (Flex) mode; b) upper ribs ON and lower ribs OFF (Hold) mode; c) upper ribs OFF and lower ribs ON (Pinch) mode; d) all ribs ON (Clamp) mode. Different configurations can provide various stiffness for the gripper’s finger to adapt the objects with different shapes and weights. To optimize the design, the stiffness analysis under various configurations and force conditions was implemented by finite element analysis (FEA). The 3-D printed prototypes were constructed to verify the feature and performance of the design concept of the VSG compared with the FEA results. The design of the VSG provides a novel idea for industrial robots and collaborative robots on adaptive grasping.

Jiaming Fu, Han Lin, I. V. S. Prathyush, Xiaotong Huang, Lianxi Zheng, Dongming Gan
Backmatter
Titel
Intelligent Robotics and Applications
herausgegeben von
Honghai Liu
Zhouping Yin
Prof. Lianqing Liu
Li Jiang
Prof. Guoying Gu
Xinyu Wu
Weihong Ren