This paper proposes a modeling of driver steering operation in lateral vehicle motion control that can be used a reference model to measure the driver intention for driver assistance system, especially aiming to derive a personalized model. In the proposed method, the control law and resulting vehicle motion obtained by applying model predictive control (MPC) to an equivalent two-wheeled vehicle model are utilized as a model representing the standard behavior of the driver. Then a model that reproduces the differences as personal characteristics is constructed and a personal modeling error corrector to reflect the differences as an addition to the MPC system is proposed. By integrating the MPC and the personal modeling error corrector, a driver steering model reflecting personal characteristics is realized. To confirm the effectiveness of the proposed method, driving experiments were conducted on M-sequence and spline interpolation trajectories using a hardware-in-the-loop driving simulator on two experiment participants, one with no driving experience and the other with 35 years of driving experience. Based on the experimental data, the proposed modeling error corrector was used to model the driver's personal characteristics and its performance was evaluated. The results demonstrated that the proposed method can model individual characteristics as a difference from the standard model based on MPC.

This paper proposes a method for designing a hybrid position/force control of a manipulator with a series elastic actuator (SEA) mounted on each joint. The actuator used in the SEA is assumed to be a low-cost servomotor with a built-in position controller. The hybrid position/force control is achieved by controlling the angle of the servomotor and the angular displacement of the SEA. Also, an algorithm is proposed to generate a target value for each of these states according to the position/force target value. To control these states, a state feedback system with an observer is designed for an extended system containing a position controller. Simulation results confirmed that hybrid control can be realized by giving the servomotor an appropriate target input according to the posture of the manipulator during the task execution. In addition, it is confirmed that the proposed control system has a certain robustness even when the initial posture and the environment parameters are different from the controller design.

This paper proposes a control method for master-slave system that considers the operating state of the human. The goal is to enable the manipulation of the robot in same manner as moving one's own arm. Since interacting with environments with different stiffness generate different level of arm tension, this paper focuses on utilizing electromyography signals from the operator's arm to estimate the arm tension with a recurrent neural network. Force control based on two types of mechanical impedance characteristics with different stiffness is constructed on the slave robot. The estimated arm tension by the recurrent neural network is used to continuously adjust the parameters of the impedance model according to the operator's intention. Experiment results show that the mechanical impedance characteristic changes according to the arm tension. When the mechanical impedance has a soft characteristic, the stiffness of the environment can be transmitted to the operator more easily. On the contrary, when the mechanical impedance has a hard characteristic, a large force can be applied to the environment. Overall, the proposed method allows the operator to arbitrarily adjust the mechanical impedance characteristics according to his/her intentions to suit the task.

This article proposes a damping control of suspended loads for crane systems that take into consideration the second bending mode oscillation occurring at the sling hook in the middle of the suspending rope. In the proposed method, a suspended load of a crane is modeled as a double pendulum, and a state feedback control based on the model is used for damping control in consideration of the second bending mode oscillation as well as the first swinging mode oscillation. The effectiveness of the proposed method was evaluated using an actual track crane. For this purpose, firstly, parameters of the model of the actual crane were identified by experiments, and the characteristics such as the valve controller and the resonance of the crane-suspended load system has been clarified to check the validity of the double pendulum model. After the preliminary experiment, a controller designed based on the identified model was implemented in the actual crane, and the performance has been compared with a conventional controller based on a single pendulum model. As the result, it has been confirmed that for both the first swinging mode and the second bending mode oscillations, the proposed controller has achieved better damping performance than that of the conventional method and that the proposed method is effective for the second bending mode oscillation that occurs in the sling hook.

Brain machine interface (BMI) is an emerging technology that aims to assist people with disabilities as well as the aged by allowing their users to intuitively control external devices by intent alone. This paper presents a signal processing technique for a low cost brain machine interface (BMI) that uses spectral analysis and artificial neural network (ANN) to classify three mental states from electroencephalographic (EEG) signals. In this study, a BMI system has been prototyped to classify the intention of moving an object up or down and at rest state. EEG signals are recorded using a consumer grade EEG acquisition device. The device is equipped with 14 electrodes but only 8 electrodes are used in this study. To evaluate the system performance, online classification experiments for three subjects are conducted. True positive and false positive rates are used as an evaluation index. Experiment results show that despite the high difficulty of the mental tasks, the proposed method is capable of achieving an overall true positive rate of up to 67% with 15 minutes of training time by a first time BMI user. Furthermore, offline analysis is carried out using the same EEG data to explore ways of using spectral analysis and ANN to reduce erroneous classifications. Analysis results show that by setting the classification threshold value higher, the false positive rate can be reduced. Another finding suggests that in contrast with the study results by other research teams, the use of multiple ANNs to classify three mental states do not improve the accuracy. Lastly, a hamming window size of 64 samples is found to be optimal for achieving real-time control when performing spectral analysis.

This paper proposes a signal processing technique to classify 17 voluntary movements from electromyographic (EMG) signals. In the proposed method, EMG signals are acquired from six EMG sensors. The features of the voluntary movements are extracted from these EMG signals using principal component analysis and later classified using artificial neural network (ANN). To evaluate the validity of the proposed method, online classification experiments are conducted on one male and one female participants. A total of 15 data sets, where each set consists of EMG signals characterizing the 17 motions, are acquired from each participant. From this total, five data sets are used as training data, while the other 10 data sets are acquired for online testing. The same experiments are repeated on a different day. The validity of the algorithm, evaluated based on the mean correct and incorrect classification rates of both days are calculated from testing data. Results show that using all five training data to train the ANN yields higher accuracy than using only one training data. The best classification result shows that there are 10 out of 17 motions with an accuracy of over 50% and mean incorrect rate of 2%. Furthermore, classification where ANN is trained using training data of a different day is also conducted. The results show that the proposed algorithm can achieve an overall correct rate of 46% at best. Based on the above results and considering the fact that users can promptly modify any erroneous actions by looking at the actual output of the prosthesis, the proposed algorithm has demonstrated the potential to classify 17 voluntary movements from 6 EMG sensors.

This paper proposes a motion design method for humanoids to imitate human-inspired motions by iteratively modifying joint reference trajectories based on the principal component analysis. The objective is to establish a practical method to realize dynamic human motions in humanoids without using complicated equations of motion. In the proposed method, captured motions of a human are firstly projected to a joint angle space of a humanoid by inverse kinematics. These joint angle trajectories are commanded to the servo motors as a function of orientation angles such a pitch angle of the humanoid, not as a function of time. Thus, the reference trajectories are commanded in a faster timing to synchronize the joint angles with the actual orientations, and this allows the humanoid to compensate the error due to the difference between the human and the humanoid in size. Furthermore, a method to iteratively modify the reference trajectories depending on resulting humanoid's motion is proposed. To extract the feature of motion, the principal component analysis is utilized. In the experiments, several motions of a human in fall were analyzed from a viewpoint of impact to the human body, and a forward-rolling motion was selected to be imitated to the humanoid. According to the proposed procedure, the forward-rolling motion was iteratively implemented to the humanoid. The experimental results showed that the humanoid achieved three quarters of the full rotation and that the proposed method could be utilized as a basis to introduce an iterative learning control to completely imitate the human-inspired forward-rolling motion.

The purpose of this study is development of high-resolution video see-through display for interactive work support. The interactive work support is a technology that can present appropriate work instructions corresponding to situations to the worker. In order to realize the technology, we are developing a video see-through display that can present work instructions. The video see-through display is composed mainly of displays and cameras. First, an image of a work object is captured by the cameras. Then, work instructions are superimposed by computer graphics (CG) on the captured image. Then, the superimposed image is displayed to the worker. Thereby, the worker can perform a task requiring an advanced skill, even if she/he is not an expert. So far, we have tried applying the technology to actual manufacturing tasks, using a prototype display system, and have performed some experiments to present the work instruction to worker. In the process of applying this technology to such manufacturing tasks, several problems have been revealed. One of the problems is that it is difficult to apply the interactive work support system to detailed tasks, because the resolution of displayed image is not sufficiently high. If the image resolution is high enough, worker can recognize the position of work object more accurately. Then, the interactive work support system can be applied to detailed tasks. In order to achieve this, we developed a full high definition monocular video see-through display. However, in principle the monocular system still had a problem, the difficulty of displaying three-dimensional position of CG. To solve this problem, in this paper, a binocular video seethrough display is developed based on monocular display. At first, the structure of binocular display is considered, and then the prototype display is constructed. In addition, some experimental results using human operators are reported.

This paper proposes a myoelectric prosthetic hand control system using a brain machine interface that uses electromyogram signals and other biosignals such electro-encephalographic signals. The proposed system consists of a feature vector extractor, a neural network estimator, a reference trajectory generator, and an adaptive corrector. As a preparation, an experimental system has been prototyped without adaptive correction using other biosignals. The system is to classify eight forearm motions by using two electromyogram signals and to generate reference trajectories for servo controllers of a prosthetic hand. The experimental results showed that the proposed classification algorithm has basically worked well and that the reference trajectory generator has a potential to make appropriate smooth trajectories. Furthermore, electroencephalogram signals during forearm motion identification were measured and analyzed to check whether reaction to indicate any malfunctions was observed.

This paper proposes a linear power assist unit to be simply designed by using an add-on end-effector and gives a controller design for the unit explicitly taking into consideration mechanical impedances of both a human operator and environment. The proposed end-effector transmits assisting force to the environment via a pair of assisting springs, and the displacement of the assisting springs is controlled to generate a desirable power assisting according to the force applied by the operator. In the proposed system, the assist gain, i.e., the ratio of the sum of the human-applied force and the assisting force to the human-applied force, can be online changed so as to easily adjust a stress loading, and it could be helpful in rehabilitations to tune the load depending on the degree of recovery. A prototype of the proposed power assist unit has been developed and was evaluated from viewpoints of both the time and frequency domains. Experimental results showed that the prototype has robustly worked well against perturbations of the environment, keeping the power assist control bandwidth about 1 to 3 Hz, and the validity of the variable assist gain control has also confirmed.

The interactive working support technology can help to perform tasks requiring advanced skills by giving appropriate working instructions interactively depending on the working situation even if a worker is not an expert. In this technology, an image presentation device is an important human interface to present the working instructions efficiently to the worker, because the image can provide a lot of information at one time. Therefore, the video see-through display is being developed as a human interface which presents the working instructions to the worker. The video see-through display consists of a camera and a display. At first, the image of a working object is captured by the camera. The working instructions will be drown by computer graphics (CG) on the captured real image. Then, the synthetic image is presented to the worker through the display. By following the presented working instructions, the worker will be able to perform the tasks efficiently. So far, we applied this technology to the manufacturing tasks, using a video see-through display we have developed, and confirmed the effectiveness of the image presentation technique we have proposed. However, there is still a problem with a low resolution when a commercially available small camera and a small display are used. In addition, the image quality is not wholly satisfactory with problems such as the distortion of the image. For this reason, it was difficult to be applied to detailed assembly tasks. That is, in order to perform more precise tasks on manufacturing site, it is desirable to develop the display system which can present higher resolution image. In this paper, the relation between the display resolution and the equivalent eyesight of presented image is discussed. And a system construction that achieves the presentation of the high-resolution image is examined. As a result of the examination, a high-resolution monocular video see-through display using full high-definition video camera and display, is developed. Furthermore, in order to confirm the basic effectiveness of the system, some results of the evaluation experiments using some operators are reported.

This paper proposes a two stage actuator system that consists of a coarse actuator driven by a ball screw with an AC motor (the first stage) and a fine actuator driven by a voice coil motor (the second stage). The proposed actuators are applied to make a wide-bandwidth bilateral control system. The first stage has a wide moving range with a narrow control bandwidth, and the second stage has a narrow moving range with a wide control bandwidth. In the proposed method, by consolidate these two inexpensive actuators based on a control design to separate bandwidth, a wide bandwidth bilateral control system can be constructed. Simulation results show that a higher transparency, i.e., better force and position responses between a master and slave, is achieved by using the proposed system compared with a narrow-bandwidth case using a single ball screw system. Furthermore, a prototype of the two stage actuator has been developed and basic performance was evaluated by experiment. Experimental results showed that a light mechanical impedance with a mass of 10 g and damping of 1.6 N/(m/s) that is an important factor to establish good haptic devices was successfully implemented.

This paper proposes an object dynamics virtualization that enables a human operator to handle the object easy for a given task by changing dynamics of the object by a robot assistance. The proposed method uses a force sensor fixed between an end effector of the manipulator and an object to measure force/torque applied by the manipulator to the object. Then, force/torque applied by an operator are estimated by using a disturbance observer, which subtracts the sensing force/torque from those calculated based on the object acceleration. Using the force sensor makes implementation easier by isolating the object dynamics from those of the manipulator. By using a virtual internal model driven by the estimated force/torque, a task-oriented mechanical impedance is given to the object that is in reality heavy and unbalanced as if it were lighter and well-balanced. Experiments to virtualize dynamics of a rod with a weight at one end, i.e., an example of an unbalanced object, to be a uniform disc by a SCARA manipulator with three degrees-of-freedom were conducted, and the results showed the validity of the proposed method.

In this paper, we propose a control method to swing up and stabilize a double Furuta pendulums (DFP).The DFP is a rotational type inverted pendulum system that has a base-link and two pendulums of different length at the both ends of the base-link. The proposed method consists of three controllers for the DFP, i.e., a swinging up controller for a long pendulum, a controller to keep a long pendulum at the upright position and to swing up a short pendulum simultaneously, and a stabilization controller for both pendulums at the upright position. Since the second controller has two objectives, we use safe manual control proposed by Astrom and Akesson. This paper proposes a novel application of the safe manual control. Swinging up and stabilization of the DFP was successfully achieved by appropriately switching three controllers. The effectiveness of the proposed method was verified by simulation and experiment.

Intelligent control of an assembly robot using vision is presented. In assembling, the task is planned by the task planner, and the corresponding robot trajectory is planned as the reference trajectory. The robot should also adapt to environmental variation. To coordinate the planner and robot, a virtual internal model is used to modify the planned reference trajectory according to the environmental variation. The use of a force and vision sensor is proposed in the virtual internal model for error correction in assembly. A simple assembling task is experimentally achieved to verify the validity of the proposed method. The task considered is screwing a bolt with an open-ended wrench fixed to the wrist of the robot, where the bolt is not accurately placed at the fixed predetermined place. The experiment shows that the proposed control system works satisfactorily.>

A single-master, multiple-slave manipulator system and its task-oriented control using VIM (virtual internal model) are proposed. By using this system, the operator no longer has to consider how to control the slave arms in coordination, so the operator can concentrate on his/her task during the operation. An experimental single-master, multislave manipulator system with two slave arms, each of which has six degrees of freedom, is described, and the concept of the system is illustrated. The experimental system does not have a bilateral function, and a master arm can specify the task-oriented variable up to six degrees of freedom. The use of bilateral feedback as well as task-oriented functions will enable the system to deal with more complex tasks.>