Published papers

© 2020 IEEE. Human motion feature extraction is necessary for robot motion generation. In particular, feature extraction methods related to non-periodic motion should be proposed. Recently, the number of the human motion recognition studies utilizing Convolutional Neural Network (CNN) is increasing due to the brilliant ability of extracting and identifying features. CNN has the same or better discrimination ability than humans so that the features extracted by CNN are thought to be useful for human motion understanding. However, since the internal structure of CNN is like a black box, it is difficult to understand the extracted features. In order to understand the features, some visualization methods which are utilized in the image field are applied in this paper. Furthermore, the many conventional studies utilizing the gradient are not preferred for visualizing CNN which recognizes human motion because human motion is sensitive for the object which treated. That is, the gradient method does not work because the value varies greatly depending on the environment even in the same motion. Therefore, new visualizing method without using the gradient is proposed in this paper. The proposed method visualizes CNN focusing part by following the neuron in CNN. Since this method does not require the gradient, the stable and accurate visualization can be performed. The effectiveness is shown by experiments.

© 2019 IEEE. Bilateral control can realize haptic feedback required for securing safety and diversity in teleoperation by robots. Since conventional approaches are based on a servo controller which has high-performance current control or torque control, there is a barrier for advancing into society. The objective of this paper is to propose a design method of the bilateral controller using velocity control driver to achieve haptic feedback. The velocity control driver such as an inverter is widely used in the industrial fields. In the proposed method, a reaction force observer modified for velocity control driver is used to estimate environmental reaction force. The control system is constructed by applying the concept of hybrid control which controls velocity and force independently, and the concept of modal space which separates motions into functions is also applied. Furthermore, the proposed bilateral controller is designed based on the ideal conditions of operationality and reproducibility. The validity of the proposal is verified through a simulation and an experiment. The obtained results show that the proposed controller successfully achieves the teleoperation with haptic feedback using velocity control driver.

© 2019 IEEE. This paper deals with the decentralized force control problem of compliant robots driven by Series Elastic Actuators (SEAs). The decentralized force control problem of a compliant robot manipulator is transformed into the robust position control problem of a servo system by using Hooke's law and lumping the nonlinear dynamics and unknown disturbances of the robot manipulator into a fictitious disturbance variable. In order to precisely follow the reference force trajectories at joint space, i.e., the position trajectories of the servo systems, the robust controller is synthesized by using Disturbance Observer (DOb) and Sliding Mode Control (SMC). Although the robust force controller can be synthesized by using only SMC, the force control signal may suffer from high control signal chattering as the force reference input is increased. In addition to improving the robustness of force control, DOb significantly suppresses the discontinuous control signal chattering by allowing to decrease the SMC gain. When the proposed decentralized robust force controller is implemented, the compliant robot can precisely track desired force trajectories and safely contact to different environments. The validity of the proposed force controller is verified by giving the simulation results of a redundant robot manipulator.

© 2019 IEEE. Bilateral control is one of the methods to realize a teleoperation with force feedback. However, communication time delay seriously deteriorates operationality and stability of bilateral control. Although 3-channel bilateral control has been proposed to improve operationality, the system becomes unstable in contact motion. Damping injection is an effective method to stabilize the contact motion by producing viscous force. However, this method produces large operational force and deteriorates operationality. Many researchers tried to inject damping without deteriorating operationality. However, their methods are affected by several situations/factors such as delay time or operator's motion speed. As a result, their methods require a redesigning procedure in order to deal with such situations/factors. In this paper, automatically-tuned damping injection is proposed to stabilize the contact motion in 3-channel bilateral control. In the proposed method, damping gain is automatically determined by monitoring the Lyapunov stability of the force control. The effectiveness of the proposed method is verified by experiments.

© 2019 IEEE. Bilateral control is getting much attention as a key technology in teleoperation. A reproducibility is one of the performance indices of bilateral teleoperation, and four-channel acceleration-based bilateral control (4ch ABC) which has a high reproducibility with wide bandwidth enables operators to feel environmental impedance, clearly. Both position control and force control are required for the 4ch ABC, and a PD controller and a P controller are often utilized with a disturbance observer (DOB) and a reaction force observer (RFOB). This paper discusses how operators can feel environmental impedance with wide bandwidth in the 4ch ABC. A damping ratio of a PD controller has been, conventionally, set about 1 by imitating a traditional design principle of servo control. This paper theoretically reveals that a velocity feedback gain of a PD controller affects a bandwidth of reproducibility as a synergy with observer gains of a DOB, and a PD controller using a high velocity feedback gain is introduced. The validity of the theoretical analysis is confirmed by simulations and experimental results.

© 2020 IEEE. In designing a human-computer interface, it is vital to understand the principle behind motion generation in humans. Previous research has mostly focused on kinematics and not kinetics. In this paper, we present the characteristics of human behavior while performing force tracking tasks. Here we present a preliminary study in which three type of measurements were conducted with three healthy young adult male subjects to characterize force modulation (discrete, rhythmic, and target transition). Subjects were instructed to control the force exerted to the load cell during a force pursuit tracking task. The visually-guided discrete attempt investigated human behavior when the target changed discretely (every 5 seconds). The rhythmic attempt investigated human behavior when the target changed continuously and periodically (every 1 second). The target transition required subjects to alter the behavior of the input force. The results suggest the following: the differential value of the force generated by a human is sufficiently large, while the response speed is based on visual information; humans can predict future reference to a small degree and attempt to improve or correct their force; and the rule of superimposition comes into effect with regard to the differential value of force.

© 2019 IEEE. In robotic systems, reduction drives are utilized to increase actuators' force. Actuators with a constant reduction ratio can not achieve characteristics of both high velocity driving and high force driving. In order to perform various tasks, high force drive and high velocity drive are required for robotic actuators since required driving characteristics are different among tasks. Therefore, a mechanism which can change a reduction ratio continuously is needed. This paper proposes a mechanism that can change a reduction ratio with a linear electro-hydrostatic actuator (EHA). The proposed mechanism switches between two discrete reduction ratios by switching two oil paths with solenoid valves. Furthermore, this paper realizes a continuously variable reduction ratio by switching the solenoid valves quickly. This paper confirms that the reduction ratio changes continuously and widely by simulations.

© 2018 IEEE. Conventional robotics have always pursued methods to bring robots near human. Human robot interaction (HRI) has gained great momentum to realize freeing robots with various control methods to ensure their safe operations with human presence. Soft robotics, alternatively, focuses on building robots mainly made of low elastic moduli materials, which may not be capable of harming humans due to nature of the materials they are made of, nevertheless, requires stiffness augmentation to be able to transmit relatively higher forces. In this study, we designed and fabricated an underactuated soft robotic finger using fused deposition modelling type 3D printing with a thermoplastic elastomer material. Robotic finger is actuated with a tendon-cable attached to a linear servo-actuator while secondary linear servo-actuator is used to control position of a stiffness-enhancement unit made of bistable metal strips. In order to verify stiffness enhancement, we perform precise position control and estimate external disturbances. Experimental results suggest that estimated external disturbances (i.e. stiffness of soft robotic finger) change under the same position conditions (finger flexion) thanks to precise robust motion controller.

© 2019 IEEE. Haptic teleoperation has been studied as an attractive technology for a long time. Although the haptic teleoperation has been already realized by conventional methods, their physical meanings are remained vague and unclear. This paper explains bilateral control for the haptic teleoperation using an equivalent circuit model. Owing to the circuit model, which takes into account characteristics of electrical actuators, a control strategy and its physical meanings are clearly illustrated. Torque (force) control and position (velocity) control for the haptic teleoperation can be designed independently because of mode decomposition. Feedback controller design for each control is based on analyses of sensitivity derived from the equivalent circuit. The control strategy derived from the equivalent circuit have been confirmed by experiments and resulted in the haptic teleoperation.

© 2018 IEEE. In this paper, a new approach for high-precise force control in repeated tasks is proposed. In the general force control, the feedback control tends to be used for a matching the output force to the desired force safely. However, when the feedback control is used, the input of the plant does not change until the error is observed. The characteristic makes reactivity bad. In the position control, the model-based feedforward control is often used to solve this problem. However, the model-based feedforward control cannot be used in force control due to the difficulty of making the accurate model. The main reason of this problem is the uncertainty of the environmental impedance. Owing to the uncertainty, the model-based feedforward control cannot work properly. To solve this problem, Iterative Learning Control (ILC) is used in this paper. ILC does not require the accurate model, and it is able to match the output force to the desired one through repeated trials. However, the convergence condition in force control has not been analyzed until now. In this paper, by taking account of the environmental impedance of the object, a new convergence restriction is proposed. Under the proposed condition, the force control in repeated tasks becomes more precise. The procedure was demonstrated on simulations.

© 2019 IEEE. This paper introduces the development of the seating posture holding assist chair and proposes the variable compliance control method. The children with cerebral palsy (CP) cannot control the seating posture themselves. In order to support their life, a wheelchair for holding the seating posture is used. However, the seating posture is holding forcibly. As a result, they get stressed and press their back to wheelchair. From the backgrounds, the seating posture holding assist chair is developed in our research. This chair has three-degrees rotational axis in reclining, twist of hip, and neck. Furthermore, the variable compliance control for holding seating posture is proposed. By experiment of reclining control, the proposed control method is verified.

© 2020 IEEE. In these days, multilevel inverters are focused thanks to their advantages, such as small voltage stress, low electromagnetic interferences, and harmonics reduction. Generally, renewable energy sources output a low DC voltage. From the viewpoint of efficiency, a single-stage conversion system which can directly change the lower DC-link voltage to higher AC voltage is required. Some single-stage conversion systems have been proposed before. However, these systems have large energy storage such as an inductor and a capacitor. This paper proposes a multilevel boost inverter stepping up the AC output peak voltage at any boost ratio by changing the number of switched capacitors. To reduce the capacitance, the proposed inverter has multiple conduction patterns when the output voltage is boosted. This characteristic distributes the discharging time to multiple capacitors. This inverter can set any maximum boost ratio and apply to the multiphase applications. The operation principle and the boost ability with multiple boost ratios were confirmed by simulations and experiments. The capacitance reduction was also confirmed by simulations.

© 2019 IEEE. Recently, for renewable energy applications, multilevel inverters are an attractive option due to their advantages, such as harmonics reduction, low electromagnetic interferences, and small voltage stress. Additionally, lower DC-link voltage should be converted to higher AC voltage, which needs an inverter and a DC-DC boost converter or a transformer. However, the two-stage conversion reduces efficiency and reliability in renewable energy applications. Hence, a single-stage conversion system which can directly change the lower DC-link voltage to higher AC voltage is required. Some single-stage conversion systems have been proposed before. However, the maximum boost ratio and the level of these systems are constant, which limits applications. This paper proposes a multilevel boost inverter stepping up the AC output peak voltage at any boost ratio by changing the number of switched capacitors. The operation principle and the boost ability with multiple boost ratios are confirmed by simulations.

© 2018 IEEE. In this paper, friction-reduced scaling bilateral control system for micro force transmission is proposed. Recently, demand for micro manipulation is drastically increasing in the fields of medical or biological science. In order to support micro manipulation, many researchers are trying to apply force feedback to the manipulation. One of the methods is bilateral control with Reaction Force Observer(RFOB). Bilateral control system is composed of a master robot and slave robot. Slave robot tracks master robot's motion and transmits reaction force to operator at master side. In the conventional bilateral control system, if reaction force is smaller than frictional force of machine structure, operator cannot distinguish the slight reaction force at hand. In this paper, actuator is supported by fiber to reduce the frictional force. Furthermore, disturbance arisen by fiber tension is modeled, and compensated. Experiments are conducted to validate the proposal.

© 2019 IEEE. Most countries are running shortage of working force due to the aging population and reduction in the birthrate. Robot manipulators are expected to replace human work. However, it it still difficult for manipulators to do simple tasks such as fruit harvesting, foods cooking or toy assembling. A problem for robotic automation arise in the difficulty in teaching how much force manipulators should use for a task execution. Motion reproduction system, which uses bilateral control to store motion data, is one of a method to teach manipulators motion including position and force. The problem concerning motion reproduction system is that the motion reproducing fails if environment is changed between motion saving phase and motion reproducing phase. Motion reproduction system which can understand and adapt to environment is required. Vision sensor can sense environment. Computer vision is mainly focus on how to classify objects. Vision information is seldom combined with motion control especially force motion. Therefore, I propose a motion reproduction system in which reproduced motion is decided based on several motions and collected depth data. Convolutional Neural Network(CNN) was used to estimate a motion command from a depth image. Saved force data was used to generate labels for training. The label decision is different from conventional Machine learning alzorithm.