Kalman-Filter-Based Sensor Integration of Variable Power Assist Control Based on Human Stiffness Estimation

In applications such as robot collaborating with human operators, the robot system must operate more slowly and be more compliant to safe user interaction. Moreover, a consideration of the dynamic properties of human operators is also important for the human application. According to such requirements, this paper presents a novel sensorless force control approach for the robot-assisted motion of the human arm. A twin direct-drive motor system with a wire rope has been developed to provide a precise force sensation and safety for human-robot interaction. In order to control the wire rope tension and human interaction force, two mode designs of the force control are realized. The common mode is utilized for the control of wire rope tension. In the differential mode, the Kalman-filter-based sensor integration for the interaction force observer is proposed in this paper. By combining two motor encoders and a commercial acceleration sensor together, white Gaussian noise is reduced, and high accurate feedback of the contact force is obtained. A variable power assist control method based on a real-time estimation of the stiffness of the human arm is also introduced. By considering the stiffness in human arm movements, this method increases the efficiency of the force control system and realizes comfortable force for human-robot interaction. The effectiveness of the method is verified by experimental results.     

Adaptive fuzzy impedance control of exoskeleton robots with electromyography-based convolutional neural networks for human intended trajectory estimation

Among various uses of exoskeleton robots, the rehabilitation of stroke patients is a more recent application. There is, however, considerable environmental uncertainty in such systems including uncertain robot dynamics, unwanted user reflexes, and, most importantly, uncertainty in user intended trajectory. Hence, it is challenging to develop transparent, stable, and wide-scale exoskeleton robots for rehabilitation. This paper proposes an adaptive fuzzy impedance controller (AFIC) and a convolutional neural network (CNN) which uses electromyographic (EMG) signals for early detection of human intention and better integration with a lower limb exoskeleton robot. Specifically, the primary purpose of the AFIC is to manage the mechanical interaction between human, robot, and environment and to deal with uncertainties in internal control parameters. CNN uses EMG signals, inertial measurement units, foot force sensing resistors, joint angular sensors, and load cells to deal with signal uncertainties and noise through automatic feature processing in order to detect user’s desired joint angles with high accuracy. EMG is particularly effective here since it reflects the human intention to move faster than the other mechanical sensors. In the experimental procedure, signals were sampled at 500 Hz as two healthy individuals walked normally at 0.3, 0.4, 0.5, and 0.6 m/s for eight minutes while wearing a robot with zero inertia. Approximately 70% of the data is used for training and 30% for testing the network. The estimated angle from the trained network is then used as the desired angle in the AFIC loop, which controls the robot online as the desired trajectory. Pearson correlation coefficient and normalized root mean square error are computed to evaluate the accuracy and robustness of the proposed angle estimation with CNN and AFIC algorithms. Experimental results show that the proposed approach successfully obtains the torque of the robot joints despite uncertainties in changing the walking speed.     

Self-reflective segmentation of human bodily motions using recurrent neural networks

For realizing a naturalistic collaboration between the human and the robot, we have to establish the intention sharing from the series of motion data that are observed and exchanged between the human and the machine. In a word, this is a problem to detect "meanings" out of the digitized data stream. In this paper, we propose a novel approach based on semiosis, and present a method of interpreting bodily motions using recurrent neural networks called Elman networks. We made some experiments using the raw data acquired while a human performs a simple task of fetching objects by stretching and folding his/her arm, and demonstrate that the network can learn invariant features of the generalized motion concepts, classify the motion by referring to self-organized memory structure, and understand a task structure of the observed human bodily motion. These capabilities are essential for machine intelligence to establishing the human-robot shared autonomy, a new style of human-machine collaboration proposed in the area of robotics.     

A nonlinear momentum observer for sensorless robot collision detection under model uncertainties

Collision detection methods could reduce collision forces and improve safety during physical human-robot interaction without additional sensing devices. However, current collision detection methods result in an unavoidable trade-off between sensitivity to collisions, peaking value reduction near the initial time, and immunity to measurement noise. In this paper, a novel nonlinear extended state momentum observer (NESMO) is proposed for detecting collisions between a robot body and human under model uncertainties based on only position and current measurements. The collision detection method is divided into three steps. The first step is to identify the robot dynamic model. Then, we can deduce the generalized momentum-based state-space equations from the identified base dynamic parameters. The second step is to construct a NESMO. Benefiting from the fractional power function and the time-varying damping ratio, the NESMO achieves the required monitoring bandwidth with noise immunity. The last step is to design a novel time-varying threshold (TVT) to distinguish the collision signal from the estimated lumped disturbance. As with the dynamic model parameters, the coefficients of TVT could be obtained by offline identification. Combined with NESMO, the method can provide timely and reliable collision detection and estimation under model uncertainties. Simulation and experimental results obtained using a 6-DOF robot manipulator illustrate the effectiveness of the proposed method.     

Interval methods for fault-tree analysis in robotics

This paper describes a novel technique, based on interval methods, for estimating reliability using fault trees. The approach encodes inherent uncertainty in the input data by modeling these data in terms of intervals. Appropriate interval arithmetic is then used to propagate the data through standard fault trees to generate output distributions which reflect the uncertainty in the input data. Through a canonical example of reliability estimation for a robot manipulator system, we show how the use of this novel interval method appreciably improves the accuracy of reliability estimates over existing approaches to the problem of uncertain input data. This method avoids the key problem of loss of uncertainty inherent in some approaches when applied to noncoherent systems. It is further shown that the method has advantages over approaches based on partial simulation of the input-data space because it can provide guaranteed bounds for the estimates in reasonable times     

Human-robot-contact-state identification based on tactile recognition

In this paper, we propose a method for designing an identification system for human-robot contact states based on tactile recognition. The following ideas are incorporated: experimentation for human-robot contact, verbalization of contact states, extraction of characteristic parameters from acquired tactile information, quantification of the recipient's tactile recognition incorporating its redundancy (identification confusability among contact states), evaluation of the identification confusability with a new criterion, and identification of contact states based on the received tactile stimulation. The proposed method allows a robot to quantify tactile recognition of a human (recipient) touched by other people (touch initiator), in which the verbal response by the recipient is matched with tactile stimulation acquired during physical contact utilizing a tactile interface. In addition, the method enables a robot that comes into contact with a human to identify contact states nearly similar to that of the recipient, based on the features of the received tactile stimulation. At this point, the reproduction of the identification confusability of the recipient's tactile recognition is also accomplished by using a neural network called modified counterpropagation (MCP). Once a tactile stimulation is induced on the robot body, the probability of corresponding contact states is calculated and outputted by the system, based on the degree of similarity of the characteristics between the newly received and previously stored tactile stimulation. Experimental results indicate that the constructed system allows a successful quantification of the recipient's contact-state recognition incorporating the identification confusability and the accomplishment of a high level of accuracy in contact-state identification. These results confirm that the proposed method is useful for identifying human-robot contact states based on tactile recognition.     

Autonomous Mobile Robot Navigation Using Passive RFID in Indoor Environment

This paper proposes an efficient method for localization and pose estimation for mobile robot navigation using passive radio-frequency identification (RFID). We assume that the robot is able to identify IC tags and measure the robot's pose based on the relation between the previous and current location according to the IC tags. However, there arises the problem of uncertainty of location due to the nature of the antenna and IC tags. In other words, an error is always present which is relative to the sensing area of the antenna. Many researches have used external sensors in order to reduce the location errors, with few researches presented involving purely RFID driven systems. Our proposed algorithm that uses only passive RFID is able to estimate the robot's location and orientation more precisely by using trigonometric functions and the IC tags' Cartesian coordinates in a regular gridlike pattern. The experimental results show that the proposed method effectively estimates both the location and the pose of a mobile robot during navigation.      

Interactive Human Intention Reading by Learning Hierarchical Behavior Knowledge Networks for Human‐Robot Interaction

For efficient interaction between humans and robots, robots should be able to understand the meaning and intention of human behaviors as well as recognize them. This paper proposes an interactive human intention reading method in which a robot develops its own knowledge about the human intention for an object. A robot needs to understand different human behavior structures for different objects. To this end, this paper proposes a hierarchical behavior knowledge network that consists of behavior nodes and directional edges between them. In addition, a human intention reading algorithm that incorporates reinforcement learning is proposed to interactively learn the hierarchical behavior knowledge networks based on context information and human feedback through human behaviors. The effectiveness of the proposed method is demonstrated through play‐based experiments between a human and a virtual teddy bear robot with two virtual objects. Experiments with multiple participants are also conducted.     

Envisioning human-robot coordination in future operations

Developers of autonomous capabilities underestimate the need for coordination with human team members when their automata are deployed into complex operational settings. Automata are brittle as literal minded agents and there is a basic asymmetry in coordinative competencies between people and automata. The new capabilities of robotic systems raise new questions about how to support coordination. This paper presents a series of issues that demand innovation to achieve human-robot coordination (HRC). These include supporting people in their roles as problem holder and as robotic handler, overcoming ambiguities in remote perception, avoiding coordination surprises by better tools to see into future robotic activities and contingencies, and responsibility in human-robot teams.     

Variable admittance time-delay control of an upper limb rehabilitation robot based on human stiffness estimation

This paper proposes a new variable admittance time-delay control strategy based on human stiffness estimation for improving the effectiveness of robot-assisted cooperative rehabilitation training. This control strategy is developed and implemented on a planar upper limb rehabilitation robot. Given the minimum-jerk-based desired trajectories of human hand position, in the developed control strategy, a time-delay approximator is utilized to estimate the external disturbances and modeling errors without exact knowledge of dynamics parameters, a sliding mode admittance controller is applied to obtained objective admittance characteristics, and an iterative optimization algorithm is used to estimate human arm stiffness and adjust human-robot interaction compliance. The closed-loop stability of the proposed control method is demonstrated via Lyapunov function theory. Experimental investigations involving ten subjects are conducted to validate the feasibility of the proposed control scheme. The experimental results show that the interaction compliance during cooperative rehabilitation training can be accurately adjusted based on selected admittance parameters and human arm stiffness, and it contributes to satisfying the specific training requirements of patients with different weakness levels and promoting the effectiveness of the robot-assisted training.     

克服噪声不确定度的拟合优度检验频谱感知算法

针对已有的基于拟合优度(GoF)检验的频谱感知算法易受到噪声不确定度影响的问题,利用矩估计法或特征分解估计法对噪声方差进行实时估计,将采样数据处理为标准正态分布的信号,最后通过GoF检验来感知主用户的存在性.在减小GoF算法复杂度的同时,克服了噪声不确定度对算法性能的影响,仿真结果也表明了所提算法的有效性.     

An Objective Function to Evaluate Performance of Human–Robot Collaboration in Target Recognition Tasks

Robotic systems in unstructured environments must cope with unknown, unpredictable, and dynamic situations. Inherent uncertainty, and limited sensor accuracy and reliability impede target recognition performance. Introducing a human operator into the system can help improve performance and simplify the robotic system. In this paper, four basic levels of collaboration were defined for human-robot collaboration in target recognition tasks. An objective function that includes operational and time costs was developed to quantify performance and determine the best collaboration level. Signal detection theory was applied to evaluate system performance. The optimal collaboration level for different cases was determined by using numerical analyses of the objective function. The findings indicate that the best system performance, the optimal values of performance measures, and the best collaboration level depend on the task, the environment, human and robot parameters, and the system characteristics. For the tested cases, the manual level was never the best collaboration level for achieving the optimal solution. The autonomous level was the best collaboration level when robot sensitivity was higher than human sensitivity. In general, collaboration of human and robot in target recognition tasks will improve upon the optimal performance of a single human detector.     

Real-time prediction of the motion tendency of human lower limbs during gait

Aiming at the problem of hysteresis in the human motion intention recognition algorithm based on kinematic sensors, a real-time prediction method about human lower limb motion tendency is proposed. It could be used to control exoskeleton robots, intelligent prosthes and other equipments in advance to eliminate the hysteresis of equipment movement. Firstly, the angle signals of ankle, knee and hip are segmented by the extreme points. Secondly, the multi-dimensional temporal association rules algorithm is used to analyze the angle signals to find out the relationships between signal patterns in adjacent time segments. Finally, the signal patterns at the next moment are predicted through the association rules algorithm, so as to predict the motion tendency of human lower limbs. Experimental results show that the proposed scheme achieves an average prediction accuracy of 78.3% for each signal segment, and can predict the subsequent motion of human lower limbs in average 92.24 ms.     

Measuring human–robot interaction on wearable robots: A distributed approach

This paper presents a novel, distributed approach to monitor physical interaction between a user and a wearable robot. We propose to apply a matrix of optoelectronic sensors embedded in a thin and compliant silicone bulk onto the user-robot contact surface. This distributed tactile sensor can measure the pressure distribution on the interaction area without affecting the comfort of the user, and does not require the robot to be specifically designed to house it. Besides the estimation of the interaction force/torque, the distributed approach allows to monitor the pressure on the user’s skin. This information is fundamental to assess the comfort and safety of the users which determine the final acceptability of the robot-mediated rehabilitation. The proposed method is preliminary evaluated on an elbow active orthosis during a repetitive rehabilitation task. Experimental results prove the relevance of this approach for the detection of the user motion intention through a measurement of the interaction force distribution.     

Human-robot contact in the safeguarding space

In this paper, we discuss a human-robot (H-R) coexistent system which allows H-R contact actions in the safeguarding space mechanically bounded by the human pain tolerance limit. The first half of this paper describes our study on the evaluation of the human pain tolerance limit which determines an individual's safeguarding space. We also show the human-safety-oriented design of a robot. The robot is covered with a viscoelastic material to achieve both impact force attenuation and contact sensitivity, keeping within the human pain tolerance limit. The robot, with simple direct-drive (DD) motor torque detection and emergency stop capabilities, automatically stops whenever any severe H-R contact occurs. In the second half of the paper, we propose a more efficient H-R system, which allows H-R contact for improving work efficiency, as long as the contact does not exceed the human pain tolerance limit. For this purpose, a robot is controlled to reduce its velocity with high reliability at an incipient stage of its contact with a human. Through experiments, we demonstrate the validity and efficient utility of the safeguarding space. The first experiment verifies that the developed robot exerts a contact force less than the human pain tolerance limit establishing the safeguarding space. The second experiment comparatively shows the robot's velocity reduction to accept a safe contact with the human in the space     

Real-Time Gait Planning for Pushing Motion of Humanoid Robot

This paper describes real-time gait planning for pushing motion of humanoid robots. This method deals with an object whose mass is not known. In order that a humanoid robot pushes an unknown object in both single support phase and double support phase, real-time gait planning for pushing the unknown object is proposed. Real-time gait planning consists of zero moment point (ZMP) modification and cycle time modification. ZMP modification is the method that modifies the influence of reaction force to ZMP. By cycle time modification, the period in double support phase is modified to avoid a robot tipping over. These modifications are calculated from reaction force on arms in every cycle. With these methods, trajectory planning for pushing an unknown object in both single support phase and double support phase is calculated. Even if parameters of an object and friction coefficient on the floor vary, the robot keeps on walking while pushing an object. The effectiveness of the proposed method is confirmed by a simulation and an experiment.     

Control and breakthrough detection of a three-axis robotic bone drilling system

This paper describes a robotic bone drilling system for applications in orthopedic surgery. The goal is to realize a three-axis robotic drilling system which can automatically stop drilling at the moment a drill breaks through bone. The proposed robotic bone drilling system consists of an inner loop fuzzy controller for robot position control, and an outer loop PD controller for feed unit force control. Moreover, breakthrough detection is a function of thrust force threshold information and trends in drill torque and feed rate. The proposed technique has been verified by drilling pig bones, the results for both the bone drilling and bone breakthrough processes are in accord with theoretical model.     

基于波束空间SRQ快速测向算法

提出一种基于波束空间SRQ算法的快速DOA估计算法,此算法通过降低相关矩阵的秩,从而节省了特征分解的计算时间。此新算法需要进行2步DOA估计,第1步用少量阵元数进行粗略的DOA估计,第2步在粗略DOA估计的范围内进行精确的估计。由计算机仿真结果可知,提出的算法大大降低了测向算法中特征分解的计算时间。因此,此新算法是一种实时DOA估计的有效算法。     

一种改进型机器人仿生认知神经网络

为了更好地模拟人类视觉系统中的注意力选择,本文提出一种改进型机器人仿生认知神经网络.首先模拟人类视觉皮层结构,在已有模型基础上建立改进型仿生认知神经网络模型;增加位置层(Position Motor,PM)到感受野(Receptive Field,RF)的自上而下(top-down)的视觉注意,同时下颞叶(Inferior Temporal,IT)不再接收全局视觉信息,而改为接收带有自下而上(bottom-up)视觉注意的局部信息,不仅降低数据处理的复杂度,也更加符合人类格式塔心理;最后利用该模型实现机器人复杂背景下目标识别与跟踪.实验结果证明该方法在有效减少数据冗余、缩短处理时间的同时,还可有效提高机器人视觉系统对目标的识别准确率.     

Development and evaluation of interactive humanoid robots

We report the development and evaluation of a new interactive humanoid robot that communicates with humans and is designed to participate in human society as a partner. A human-like body will provide an abundance of nonverbal information and enable us to smoothly communicate with the robot. To achieve this, we developed a humanoid robot that autonomously interacts with humans by speaking and gesturing. Interaction achieved through a large number of interactive behaviors, which are developed by using a visualizing tool for understanding the developed complex system. Each interactive behavior is designed by using knowledge obtained through cognitive experiments and implemented by using situated recognition. The robot is used as a testbed for studying embodied communication. Our strategy is to analyze human-robot interaction in terms of body movements using a motion-capturing system that allows us to measure the body movements in detail. We performed experiments to compare the body movements with subjective evaluation based on a psychological method. The results reveal the importance of well-coordinated behaviors as well as the performance of the developed interactive behaviors and suggest a new analytical approach to human-robot interaction.