Robotic power assistance using surface electromyography
   Power assistive devices that employ SEMG signals can provide mechanical power to supplement human physical strength according to the estimated torque from the measured muscle activation. It also allows us to effectively support the user in his/her participation in the rehabilitation program by extracting the movement intents of voluntary exercise from SEMG. The assistive devices are expected to be useful solution for individuals who have deficits in motor functions but who still have measurable SEMG, as well as labors and servicemen; therefore, many researchers have recently been working on SEMG signals to control the assistive robotic devices that will assist users.
   Our research have been focusing on the evaluation of movement instability by power assistance, which prevents interfaces from lending natural and seamless assistance. The devices are inherently a feed-forward system, and this characteristic can hinder natural and seamless assistance because of the noisy nature of SEMG signals. The upper limit of the amount of torque assistance and the movement instability by the amount have been quantifying; such an evaluation of these devices can lead to refinements that will result from assist characterization, The outcome of this study can help researchers to refine assistive using assist characterization, and can be helpful for the physical therapists who have to tune such devices based on quantitative standards and their own experience.
Figure 1 SEMG-based elbow flexion assistance of a stroke survivor (right hemiplegia)

[Figure 1] SEMG-based elbow flexion assistance of a stroke survivor (right hemiplegia)


Wearable ground reaction force measurement system
   A stationary force plate is the standard method to measure the ground reaction force (GRF) in a laboratory environment. Since the force plate is spatially confined to a laboratory space, measuring the GRF data under various conditions, such as climbing up stairs and jumping and walking outdoors, is a challenging task. We proposes a compact GRF meausrement system equipped with high resolution in-house force sensors. The use of optical based sensor which does not require additional signal amplifiers makes the system compact. The system was designed as outsole type platform separated into three parts (toe, ball, and heel part) to reduce interference and adjust to the foot curvature. Seven in-house force sensors were used; two sensors (x and z direction) at the toe, three sensors (two z and one y direction) at the ball, and two sensors (y and z directions) at the heel. The measured forces were transmitted to host computer by an embedded module nearby the platform. The calibration of the proposed system was undertaken using a commercial force sensor.
   The proposed system which has mobility and comparable performance to commercial system can be used in biomechanics and rehabilitation where spatial constraints often become a barrier to the analysis of the GRF.

[Figure 2] Wearable ground reaction force measurement system


Body-powered orthosis using hydraulic cylinders
   To realize a rehabilitation robot that is untethered and wearable for long periods, we propose an architecture based on transmissions rather than motors. The robotic appraches in rehabilitation field have been focused on the realization of the perfect human movement having mult DOF for wearable robotic system. However, the system becomes heavy and complex to user without assistant. In addition, the cost of the system is burden to the clinical facilities as well as patients. As an alternative of the drawbacks of the systems, we propose a body-powered robotic system. Power and control would be derived from a patient’s unimpaired or less impaired limb functions and routed through a transmission to drive impaired limb functions. Because the device is body-powered, the patient is always in control and always provided with force feedback. To determine the feasibility of such an approach, we are prototyping variable hydraulic transmissions using off-the-shelf cylinders.

Powered exoskeletons for grip force assistance
   Hand grip capability plays a crucial role in wide range of activities of daily living (ADL) and physical works. For those who have weakened grip strength due to diseases, grip force assistance devices can help performing successful activities by providing additional grip force. In addition, people who exert repetitive and large grip force, particularly if combined, are exposed to high possibility of work-related musculoskeletal disorders (WMSDs) including carpal tunnel syndrome. Because relieving hand grip force is important in avoiding WMSDs, grip force assistance devices can help them prevent WMSDs by providing controlled large force which is an amplified version of human grip force.
   Powered exoskeletons for grip force assistance have been developed by several research groups over the world. However, because the majority of them use finger contact force as a control signal, the existence of force sensor at contact surface prevents the fingerpad from being directly exposed to environment. Therefore the wearer’s tactile sensation at fingerpad is inevitably distorted. This tactile obstruction reduces the manipulation dexterity because human performance of grip force control largely relies on the tactile sensation during manipulative tasks.
   Our research points to the development of a hand exoskeleton with partially open fingerpads. It allows direct contact of the the wearer’s fingerpad and environment so that the tactile sensation is preserved. Meanwhile, the grip force given by the wearer is estimated using the measurement of partial contact forces at both sides of the finger. Actuators are then controlled to exert desired assistive grip force which is determined based on the estimated grip force.

[Figure 3] Partially open hand exoskeleton for grip force assistance

Soft and stretchable strain sensors
   Soft strain sensor is one of the major interests for mechatronics, human-machine interface and wearable technology since it can be utilized in soft robots, human-robot interactive devices, artificial skins, and many other related systems. Although a number of studies on the highly sensitive and stretchable soft strain sensors have been published, strain sensors covering large area and having arbitrary three dimensional shapes have not been well explored. Especially, very few studies have focused on the soft strain sensors that can simultaneously acquire strain distributions at multiple locations over large and complex shaped surfaces without fabricating complex array of stretchable electrodes within the sensor body.
   We suggest multi-point and multi-directional strain mapping sensor by using soft nanocomposites and a computational approach based on electrical impedance tomography(EIT). The soft nanocomposites were fabricated using multi-walled carbon nanotubes(MWCNT) and silicone-based elastomer. As an advantage, this approach can be made in various three dimensional shapes with low manufacturing cost. Furthermore, the sensor can work reliably even with harsh contact conditions at the sensing surfaces because the electrodes can be located away from the sensing surfaces.

[Figure 4] Demonstrations of 3-D shaped soft tactile interface

Magnetic resonance (MR)-compatible hand device
Due to the high spatial resolution provided by functional magnetic resonance imaging (fMRI) technology, a large amount of research has been conducted using fMRI, such as the investigation of brain activity during functional activation tasks of the hand, including making a fist, finger flexion/extension, finger tapping, and tracking of pre-defined trajectories. However, without utilizing mechatronic systems composed of sensors and actuators, it has been a challenge to guarantee high repeatability over repeating test conditions and between subjects.
We developed a two degree-of-freedom (DoF) magnetic resonance (MR)-compatible hand device that can provide such functional activation tasks and robotic rehabilitation procedures inside an fMRI-scanner. The device is capable of providing real-time monitoring of the joint angle, angular velocity, and joint force produced by the metacarpophalangeal (MCP) and proximal interphalangeal (PIP) joints of four fingers. For force measurement, a custom reflective optical force sensor was developed and characterized in terms of accuracy error, hysteresis, and repeatability in the MR environment. The proposed device also consists of two non-magnetic ultrasonic motors to provide assistive and resistive forces to the MCP and PIP joints. With actuation and sensing capabilities, both non-voluntary-passive movements and active-voluntary movements can be investigated with the proposed system.
mrdevice1 mrdevice2

[Figure 5] A structure and components of magnetic resonance-compatible hand device