Search

With the development of robotic technology, the demand for state-of-the-art technology in the field of rehabilitation is rapidly increasing for the elderly and people with disabilities. In this paper, we propose a real-time virtual coach to assist physical therapists with the end-effector-based robot-assisted gait training for stroke survivors using Long Short-Term Memory (LSTM) networks. Our proposed virtual coach consists of the sensor module for data gathering and dataset generation, real-time classification of the pathologic patient gait during the training using LSTM networks, and delivery of the coaching recommendations in an audiovisual form. Our preliminary study determined the selection of coaching recommendations. LSTM networks are trained to provide the selected coaching recommendations. The performance of the proposed virtual coach is verified using classification simulation of an able-bodied person on the rehabilitation robot, G-EO System. The usability was verified through a satisfaction survey of five professional physical therapists.

Use of end-effector type robots might be more effective to gait rehabilitation than use of exoskeleton type robots, but there has been little work-in-progress to investigate gait pattern when using end-effector type robots. In this study, wearable inertial measurement units (IMU) signals of body segment during walking with the end-effector type robot were compared with during the walking on the treadmill. Different characteristics during the walking with the two configurations were observed; small range of angular velocity and no impact acceleration on the heel-strike during walking with the robot were shown. Since the foot of the subject was fixed on the footplate (i.e., end-effector) of the robot and footplate trajectory of the robot indirectly generates joint motions of lower limbs, the trajectory should be carefully designed. Therefore, the results are applicable to identify the desirable and effective trajectory of an end-effector type gait rehabilitation robot.

The NRC Robotic Exoskeleton (NREX), an upper-extremity, rehabilitation robotic exoskeleton developed for daily life exercises, and treatment of hemiplegic or spinal-cord injured people, was improved by the National Rehabilitation Center, Korea. The shoulder module, which had restricted movement, was improved via additional passive-axis movements with rubber bands. In addition, NREX can now be used for either left- or right-arms by changing only two parts of the robot arm. We focus on the improvement of NREX mechanism to reduce cost and improve usability.

Two-dimensional strain in a nonpolar InGaN/GaN quantum well was measured quantitatively using inline electron holography. A periodic undulation of the strain was observed to arise to compensate otherwise diverging potential associated with the in-plane polarization.

In therapeutic rehabilitation devices, different types of feedback are provided to improve the outcome of the rehabilitation process. Lower limb robotic exoskeleton devices for overground gait rehabilitation have difficulties providing subjects certain types of feedback, such as visual feedback, because of the restriction of the visual display installation space and the mobility of the whole system. As a remedy, we propose the concept of providing auditory stimulation with force feedback via exoskeleton legs. As a preliminary study, we implemented an Android OS-based rhythmic auditory stimulation module in order to effectively connect exoskeleton robotic devices. Two 3D accelerometers were attached around the subjects' ankles. By using this module, we performed a preliminary study on rhythmic auditory stimulation with able-bodied participants and one stroke participant. We observed changes in gait speed, cadence, and stride length in accordance with the rhythmic auditory stimulation. Based on these findings, a smartphone-based rhythmic auditory stimulator could add value to exoskeleton robotic devices.

Various robotic systems have been proposed to provide assistance in activities related to daily living and in physical therapy. Our consortium developed a prototype assistive robotic module, called the "smart mobile walker," which is a motorized walker. This robotic device was developed to assist with the mobility of senior citizens and persons with disabilities such as stroke survivors and people with spinal cord injuries. A pilot study was performed with four senior citizens to evaluate the usability of the system relative to a commercial walker. The participants performed a "10-meter walk test" (10MWT) with either the smart mobile walker or a commercial walker on flat, upward-sloping, and downward-sloping surfaces, indoors over five days. An electromyography (EMG) signal and the heart rate were recorded, and the gait parameters including the speed, cadence, stride length, and gait cycle duration were measured to evaluate the system usability. The participants performed better with the commercial walker on the flat and upward-sloping surfaces owing to the unit's light weight. On the downward-sloping surface, however, the speed control offered by the smart mobile walker was found to outweigh the advantage of the light weight of the commercial walker.

Various robotic systems have been proposed for providing assistance in activities of daily living and support therapy. Our consortium simultaneously had developed an assistive robotic module, called a smart mobile walker, and its application scenarios on the basis of inputs obtained from candidate users and clinical experts during the entire robot development process. This paper presents the development of scenarios for examining the applicability of the robotic module in providing support for activities of daily living. Scenario construction is an essential process for embodying the target population, purpose, and performance of an assistive robotic system in its development process. To this end, in this study, we conducted an in-depth analysis of the feedback received during interviews of a focus group and clinical experts and consulted the International Classification of Functioning, Disability and Health. Finally, we developed eight scenarios for the smart mobile walker that were targeted at elderly people and stroke survivors.

This paper reports the development of a prototype of a novel exoskeleton robot, i.e., National Rehabilitation Center Robotic Exoskeleton, called NREX, which was designed to assist individuals who have sustained neurological impairments such as from stroke and spinal cord injuries. Among the activities of daily living, we focus on reaching and drinking tasks using an upper extremity. For practicality, we only adopt two motors for the wrist and elbow joints, while using passive devices for the other joints. In particular, the wrist motor can manually be adjusted along the internal/external rotation motion for the simple implementation of activities of daily living tasks.

We present a novel immersive upper extremity exercise testbed for activities of daily living (ADLs). The system comprises a seven degrees-of-freedom robotic arm and an off-the-shelf projective 3D display. The manipulator and the display device correspond to the motor function and cognitive function, respectively. ADL tasks that involve upper extremity exercises are important for improving real-life functional performance. ADL tasks usually involve manipulation and/or reaching for an object between desk level and head level. Of the various movements possible, our participants performed point-to-point reaching movements, which are related to ADL tasks, using the proposed system, and we compared 2D and 3D display devices. Three groups, i.e., able-bodied, spinal cord injury, and chronic stroke groups, participated in the reaching tasks. The 3D display was more effective for performing point-to-point reaching movements.

Various robotic systems have been proposed to assist in daily living and support therapy. Even though several assessment tools and standards exist for assistive devices, medical devices, and industrial robots, most assessment tools are not suitable for application to novel conceptual systems. In particular, most developers have difficulty in assessing the developed robotic systems in the rehabilitation domain. We introduce a newly designed assessment tool for smart mobile walkers-that is, robotic devices that assist the mobility of senior citizens and people with disabilities, such as stroke survivors, and people with spinal cord injuries.

Innovative assistive robotic modules are developed based on the inputs obtained from the user candidates and clinical experts during the entire robot development process. This paper presents scenarios developed for examining applications of activities of daily living support robotic modules. We conducted an in-depth analysis of the feedback received from the focus group and clinical expert interviews and constructed 16 scenarios for an end-effector-based upper-limb robot, an exoskeleton upper-limb robot, a smart mobile walker, and a low-limb exoskeleton robot. The major candidate populations for using the above robotic systems are elderly people, stroke survivors, people with spinal cord injuries, and people with myopathy.

Reaching is one of crucial tasks in order to assist activities of daily living and to support therapeutic rehabilitation. We investigated the dynamics of upper-extremity motions for the reaching tasks at different levels by stroke and able-bodied participants. The two different levels were at the head and chest. The results indicated that the joint moments with respect to levels differed slightly for the stroke and able-bodied participants. The stroke participants showed an increase in the shoulder flexion-extension moment compared to the able-bodied participants for reaching tasks at the upper level. Stroke participants showed an increase in the elbow flexion-extension moment compared to able-bodied participants for reaching tasks at the lower level. The dynamics results of the motions at different levels may provide a fundamental understanding of the design for assistive robots to help stroke survivors with activities of daily living tasks including vertical motions. To be able to support stroke participants by using assistive robots for ADL tasks, the shoulder flexion-extension motion should be strongly supported for vertical movements of an upper extremity. For therapeutic application, the abnormally large joint moments of stroke participants can be restricted.

A mobile platform-based rehabilitation robotic system with a lower-extremity exoskeleton was developed for people with disabilities to allow them to perform therapeutic exercises and to support daily living activities. The system consists of a weight support module, a locomotion module, and an exoskeleton module for lower extremities. We collected the needs and feedback of users and clinical experts from a previously developed system. In this paper, we review important points for the design of a system for people with motor impairments of the lower extremities. This system is an example of a human-in-the-loop system. We focused on the physical interaction between the human subject and system. As a pilot study, we checked the pressure of the harness supporting the weight of the subject.

The function of lower-extremity rehabilitation robotics usually includes standing up movements. These standing up movements can cause spasticity in people with disabilities of the lower extremities. In this paper, we review the spasticity and the methods of existing systems for handling the spasticity of people with physical disabilities. We observe and detect the spasticity via the EMG signals of the lower extremities of a subject with paraplegia. Through this process, we are able to suggest ways to deal with risk factors, such as spasticity, when using rehabilitation robots.

A novel lower-extremity robotic exercise system with weight bearing and locomotion abilities has been developed for the severely disabled. The purpose of this study was to investigate the needs of users and clinical experts when using this robotic exercise system. The system consists of an exoskeleton, weight-bearing system, interface, and system platform. The results of this research are based on the opinions of focus groups and clinical experts who were interviewed after applying the robotic system. Comfort, safety, and natural appearance while wearing the system were considered. User feedback, received from the initial design and development of our system, has contributed to the implementation of our rehabilitation robot.

Walking and sit-to-stand support system, the smart mobile walker, is being developed to help elderly and disabled to live an independent daily life. It comprises almost omnidirectional driving mechanism, sit-to-stand support mechanism, and motion compliance module to control the system. The development and evaluation of basic motion compliance algorithm shows that the algorithm for extraction of user intention from the interaction of the operator and the system determines performance of the system. The required parameters and performance index for improving sit-to-stand function are evaluated by the analysis of force reflection between the operator and the system. Finally, future research topics to improve the performance are addressed.

The topic of enterprise architecture (EA) has been gaining significant attention from both academia and industry due to the inefficiency of current IT architectures to cope with rapid changes in business environment. In order to turn existing EA into efficient and agile one, it is necessary to institutionalize an EA based on well-established enterprise architecture frameworks (EAFs) into the enterprise. In this paper, we propose EA institutionalization processes and its metric based assessment for implemented EA based on the currently available EA frameworks. In the EA processes, we define institutionalization strategies specific to organizations’ goals, target architecture based on their baseline architecture, and transition plan for institutionalization.

In order for a software system to be more beneficialto an organization, its services should be laid out in a carefulconsideration of the various constituents of the organization, orthe architecture of the organization. Here after enterprisearchitecture. A software system and the enterprise architecturethen ultimately help achieve the goals that the organizationhas, while helping to solve some critical problems theorganization might be faced with. In this paper, we present ourinitial understanding of the role of an enterprise architecturein relation to such goals and problems, as well as to therequirements on, and the specification of, a software system.Based on such understanding, we then present our preliminarywork on how to proceed towards institutionalizing anenterprise architecture for an organization, with an emphasison the high-level concerns that an enterprise architect shouldhave during the process of institutionalization.

This paper describes a method for measuring the efficiency of a sequence of each layer in layered pattern recognition. Considering pattern recognition as nonlinear transforms of input patterns, we claim that the efficient layers reduce variations within the classes while maintaining separability among the classes. In order to quantify these values, we define two measures, named as inter-class separability and intra-class variation, based on entropy. We propose a method to analyze the efficiency of pattern recognition layers using these measures. Applying this method to a handwritten digit recognition system, we can successfully identify inefficient layers.

KARES (KAIST Rehabilitation Engineering System) is a rehabilitation robotic system with 6 degrees of freedom robot arm mounted on the powered wheelchair, in order to assist the disabled and the elderly for independent livelihood. Human machine interaction is very important in this system. The direct control of the robot arm takes a high cognitive load on the user's part while physically disabled persons may have difficulties operating joysticks or push buttons for delicate movements. Therefore, a certain degree of autonomy of the robotic system is needed. To perceive the environment, color vision and force/torque sensors are mounted on the end-effector of the robotic arm of KARES. Four basic tasks are tested: picking up a cup from a table, picking up a pen from the floor, moving an object to the user's face, and operating a switch on a wall. These tasks are performed autonomously in the semi-structured environment.

Human-robot interaction technologies are very important components in the human centered system. Among many human-robot interaction technologies, the intention reading of a user plays an important role in human centered systems. In the paper, we focus on the use of intention reading in rehabilitation robots, and implement its capability for the wheelchair-based robotic arm system, called KARES (KAIST Rehabilitation Engineering Service System) II. For effective intention reading, visual servoing by using an image sequence of user's face or control by using EMG signals of user's arm are applied to KARES II. Specially, it is shown that soft computing techniques are very effective for intention reading.

Intention reading is one of the essential functions of human-friendly rehabilitation robots. First, the overall structure of a new wheelchair-based robotic arm system, KARES II, and its human-robot interaction technologies are presented. Among those technologies, we concentrate on visual servoing that allows this robotic arm to operate autonomously via visual feedback. Effective intention reading, such as recognizing the positive and negative meaning of the user, is performed on the basis of changes of the facial expression around lips that is strongly related to the user's intention while serving the user. For the efficient visual information processing, log-polar mapped images are used to control the stereo camera head that is located in the end-effector of the robotic arm. The visual servoing with effective intention reading was successfully applied to serve a beverage for the user.

Visual servoing gives autonomous capabilities to a rehabilitation robot for human-robot interaction. Though several visual-based control techniques have been proposed in rehabilitation systems, it is not easy to get the accurate visual information of a target, i.e., depth, in real-time. Specially, the depth information of a target, such as the distance between a user and a robotic arm, is closely related to a user's safety. As a remedy for this situation, visual servoing based on space variant vision is proposed for an eye-in-hand camera configuration. Space variant vision with vergence movements of a binocular camera head can provides well a robotic arm with depth information. In addition, a small and light-weighted camera head is designed and implemented.

We provide an introduction to the overall structure of a new wheelchair-based robotic arm system, KARES II, and its human-robot interaction technologies which assist independent life of the elderly and the disabled that have disadvantages in sensory and motor functions of their limbs. First of all, the tasks of a robotic arm are specified. Next, several human-robot interaction technologies are presented to perform the defined tasks in comfortable and safe ways. In addition, the input/output relation of several interaction technologies, such as an eye-mouse, a haptic suit, bio-signal recognition and visual servoing, is described under two essential factors: one is intention reading of the user, and the other is the autonomous capability of the robot. Among those technologies, we concentrate on visual servoing that is a good solution for reducing the burden of a user in a wheelchair with a robotic arm. We mention difficulties of visual servoing in the wheelchair-based robotic arm. It is found that visual servoing using space variant vision has several merits in an eye-in-hand configuration. Specially, a nearby target is handled efficiently through a modified log-polar mapping.

A high quality deflection yoke (DY) is an essential factor for a high quality monitor. Currently, the DY adjustment process is done manually by attaching ferrite sheets on the inside surface of the DY. In this paper, we deal with the convergence adjustment algorithms employed in the guidance system. An inference engine that is based on soft computing techniques is proposed to systematically deal with several resources including the expert's knowledge in the convergence adjustment process of the DY. In our approach, the rough set theory is used to handle the input/output data collections of the DY adjustment, and the fuzzy logic and evolutionary programming are used to model the plant and to tune the model. With the given initial convergence, a rough set driven coarse search part finds a coarse area and then more precise positions are searched and decided in the fuzzy inference engine driven fine search part. Initial experimental results show that the proposed algorithm works well in the real convergence adjustment process of the DY.

An efficient method which yields an appropriate number of hidden nodes by iteratively removing unnecessary hidden nodes is proposed. The criterion for selecting the removable nodes is derived from the analysis of information transformation of individual neurons and the network as a whole. A well-trained multilayer perception (MLP) that has a sufficient number of hidden nodes is used. A hidden node is selected with an output weight of the smallest magnitude, and it is removed from the hidden layer. After the removal, additional training is performed. This procedure is iterated until the network performance is not affected severely. The method has been applied to some problems, including handwritten digit recognition, resulting in a significant reduction of hidden nodes. >

The classification performance after a single iteration is derived as a function of the number of exemplar patterns, the number of processing elements in the network, input bit error probability, and inner-product values among exemplar patterns. It is shown that, when the exemplar patterns are orthogonal with each other, the correlations among the weighted sums are nothing but constant multiples of the weights. These values are rarely zero, which makes the analysis complex. However, the values of correlation reduce greatly when these weighted sums pass through the threshold. The classification performances after single and multiple iterations are derived using this method and compared with simulation results. The results can be a valuable reference in the design of Hopfield-type associative memories