To develop a nonverbal communication channel between an operator and a system, we built a tracking system called the Adaptive Visual Attentive Tracker (AVAT) to track and zoom in to the operator's behavioral sequence which represents his/her intention. In our system, hidden Markov models (HMMs) first roughly model the gesture pattern. Then, the state transition probabilities in HMMs are used to assign as the rewards in temporal difference (TD) learning. Later, the TD learning method is utilized to adjust the action model of the tracker for its situated behaviors in the tracking task. Identification of the hand sign gesture context through wavelet analysis autonomously provides a reward value for optimizing AVAT's action patterns. Experimental results of tracking the operator's hand sign action sequences during her natural walking motion with higher accuracy are shown which demonstrate the effectiveness of the proposed HMM-based TD learning algorithm of AVAT. During TD learning experiments, the exploring randomly chosen actions sometimes exceed the predefined state area, and thus involuntarily enlarge the domain of states. We describe a method utilizing HMMs with continuous observation distribution to detect whether the state would be split to make a new state. The generation of new states brings the ability of enlarging the predefined area of states.

The goal of our study is development of a "tactile mirror" for displaying tactile perception information. A point of distinctive difference between a tactile mirror and conventional tactile displays is that a tactile mirror possesses a feedback mechanism for acquiring the contact state information on the finger pad concerning not force but tactile perception sensation. The primary development concerns human static friction sensation. First, a control method and the structure of a tactile mirror are described. Second, simulation results using FEM for contact analysis and a ANN(artificial neural network) for generating feedback signals are shown. Finally, experiments demonstrating that providing a feedback component accomplish accurate display of static friction sensation with a conventional type of tactile displays.

It is considered that kinesthetic feeling induced by the kinetic characteristic of a manipulated object plays an important role for performance of human daily skills. Therefore, if we consider a wearable robot/joint as the robot/joint operated by the daily skills, it is necessary to control the robot/joint so that kinesthetic feeling induced by the kinetic characteristic of an object is directly transmitted to a wearer, without any distortion by the kinetic characteristic of the wearable robot/joint itself. This is the control issue of "kinesthetic transparency". This paper discusses a method of control to improve kinesthetic transparency of a wearable joint. Kinesthetic transparency can be basically improved by active compensation. In order to avoid necessity of offline estimation for unknown kinetic parameters of the joint, this paper formulates a control law which automatically estimates the unknown parameters during operation of the joint. Simulation results demonstrate effectiveness of the method. However, experimental results reveal that accurate compensation of frictional terms is necessary for further improvement.

To successfully incorporate human skills into a wearable robot, it is required to design its joint mechanism so that it does not interfere with the human skills. From this viewpoint, this paper first addresses six design goals for the joint mechanism. Based on the result of preliminary examination for two well-known mechanisms, a new 6-DOF mechanism referred to as the Wearable HEXA mechanism is proposed. This is based on the HEXA, which has been originally developed as a high-speed manipulator. Detail of its design process is given, and evaluation of a trial mechanism manufactured along the design process is also reported.

We propose a method for preventing hazardous accidents due to operator's action slip in their use of skill-assist, which has been already introduced as a power assist device in automobile assembly processes. First, we show that hidden Markov model (HMM) can be used to detect human erroneous operation from data sequences of an operator's hand motion trajectory, but that a problem arises in direct application of HMM to human error detection when trajectory is not in alignment with any of pretaught trajectories. However, HMM-based Dempster-Shafer theory proposed in the study allows the system to select their safety- or productivity- oriented robot control policy, solves the detection problem, and performs secure accident prevention, Second, a workability improvement process which comprises a state-policy and a teach data renewal sub-processes allows for optimal reconstruction of the policy determinant state space and HMMs. Finally, experimental results verify our overall proposal, and demonstrate the effectiveness of the workability improvement process.

The study concerns a human error recovery function for efficient human/robot collaborative parts conveyance tasks, which has not been discussed so far in construction of human/robot coexistence systems. We propose a human operational intention inference updater which allows a robot to judge whether the goal intended to reach by the operator is correct or not, from incomplete trajectory data sequence of the robot endtip maneuvered by the operator. Experimental results showed that a use of the proposed inference updater reduced the task execution time even when human error occurred in a task, which demonstrate the usefulness of the updater.

Recent advances in vision technology lead to the establishment of non-verbal communication channels from human behaviors toward understanding their intention. An adaptive vision-based attentive tracker (AVAT) is proposed for isolating such human intended actions from the ordinary walking behavior. The algorithm which drives the attentive tracker is divided into two sub-processes: one is for modeling the movement of human body parts as the environment using hidden Markov models, and the other is for learning the model of the tracker's action using a temporal difference algorithm. In the paper, we describe in detail the integration of the two sub-processes and finally show an experimental result of isolating the human sign action during his natural walking motion for demonstrating the feasibility of our proposal. Identification of the sign context determines the action of the tracker with the simulated rewards supplied.

It has been pointed out that the hard structure of an EVA glove deteriorates efficiency of tasks in the space environment. We also found a claim that an EVA glove did not allow an astronaut to acquire contact information at the fingertips. In the study, we proposed a SkilMate Hand for space EVA gloves which has both a tactile media and a power assist devices. We locate SkilMate in a wider framework of wearable intelligent machines which assist in affording such working surroundings that they can exhibit their skills in spite of their necessity for wearing special suits typically in hazardous environments. The paper focuses on the development of a finger system of a SkilMate Hand, and begins with determination of the design guidelines through analyses from an interview with an astronaut. To improve the task efficiency, we manufacture a power assist device which compensates the bending moment exerted at a human finger joint utilizing an ultrasonic motor. We show the structure and the control strategy of the motor drives. To overcome the second problem of tactile insensibleness, we produce a tactile media device which is composed of a slip sensor element on the outer side and a vibrotactile display element on the inner side of a SkilMate Hand. Transduction characteristics of both pressure sensitive conductive rubber and piezo-rubber are experimentally examined in the low temperature region. Finally, piezo-rubber is chosen to be used as a slip sensing transducer.

We propose a human-will-based approach to constructing a coexistence system in which a robot is assumed to assist the will power of a subject while reflecting his intention and desire. We discuss the model of human's will associated by the work - psychological mind reading, and identify the advantages of the system in which the robot guides the the subject manipulating along the path intended by him. Then, we describe two major techniques with experimental results: 1) estimation of the human-intended path pattern from the previously assigned one by proposing a modified algorithm of hidden Markov model; and 2) merging of the human's force feeling pattern into the desired one by proposing a field impedance equalizer which allows both the human to express his desirable field-dependent force/speed feeling and the robot to display the impedance along the previously estimated path. Finally, we conclude the study with some comments on future works.

This paper mainly concerns a vibrotactile sensor capable of generating impulsive high-frequency signals by which only slipping states of an object can be distinguished. Slip information is so indispensable to guaranteeing the grasp of an object and detecting static friction coefficient that various kinds of slip sensors have been proposed. However, there has been no report on such a slip sensor that slip signals supply real-time critical information to extract only a slipping state from various contact states. We propose a sensor structure with surface ridges in the concavity of which strip-shaped sensor elements are vertically fixed. We analyzed the slip signals of the sensor and showed that the output signal in slipping states was originated in the free vibration mode of the sensor elements. Then, we verified, by using a two-fingered robot hand that the slip signals of the sensor could distinguish the slipping state from other contact, release, or rolling state. Finally, we demonstrate that the slip signal of the sensor output guarantees a slipping state to increase the grasping force of the two opposing robot fingers not to drop the object, by making full use of the impulsiveness of the slip signals.

Autonomous distributed control (ADC), as a promising approach in dealing with complex control problems, is capturing the attention of automation specialists, yet there is still no available platform supporting its industrial application. This paper introduces an ongoing project toward this goal, carried out in our laboratory. In this work, we use agents and an agent society to model an ADC system. We propose a new agent architecture for intelligent machines and a supervisor agent for humans to provide supervision control for agents. To facilitate the coordinating behaviors of agents, an agent communication language called CCL (Cooperative Command Language) is developed, and an object-oriented implementation of the proposed agent system is presented, too. Furthermore, the support system under development is introduced in a simple way. A case study concerning the shape control of a variable-geometry truss is also given to show the availability of the agent system.

A major intestinal movement known as peristalsis was studied as an autonomous distributed system. Although the small intestine takes a homogeneous structure along the longitudinal axis, the intestinal movement is autonomically systematized to transport the contents smoothly by itself. To study control mechanisms, a dog's small intestinal movements were observed in vivo, and two types of peristaltic movements after feeding were found. The transport efficiency of peristalsis was analyzed, using a mathematical model. The numerical analyses showed that there is an optimal movement that maximizes the transport efficiency. Neurodynamical simulations with mechanical peristaltic simulators that can handle actual materials were also studied. The proposed neuromechanical model transported various kinds of contents, varying the movements to suit the properties of the contents.>

The authors present a basic formulation of motion dynamics of a free-floating rigid-link system to establish a basis of the collision dynamics. They propose a novel concept named extended generalized inertia tensor (Ex-GIT), which is an extended version of the GIT for ground-based arms, and discuss the virtual mass concept. By means of these concepts, they formulate the collision problem focusing on a velocity relationship just before and after the collision without sensing the impact force, but considering the momentum conservation law.>

An analysis of the momentum conservation equations of a redundant free-flying spacecraft/manipulator system acting in a zero-gravity environment is presented. In order to follow a predefined end-effector path, the inverse kinematics at velocity level is considered. The redundancy is solved alternatively in terms of pseudoinverses and null-space components of the manipulator inertia matrix, the manipulator Jacobian matrix, and the generalized Jacobian matrix. A general manipulation task is defined as end-effector continuous path tracking with simultaneous attitude control of the spacecraft. Three subtasks of the general task are considered. The case of manipulator motions that yield no spacecraft attitude disturbance is analyzed in more detail and a special 'fixed-attitude-restricted' (FAR) Jacobian is defined. Through singular-value decomposition of this Jacobian, corresponding FAR dexterity measures (FAR manipulability and FAR condition number) are derived.>

The control problem of multiple manipulators installed on a free-flying space robot is presented. Kinematics and dynamics are studied and the generalized Jacobian matrix is formulated for the motion control of a multiarm system. Individual and coordinated control of dual manipulators is discussed. For the coordinated operation, a new method of controlling two arms simultaneously-one arm traces a given path, while the other arm works both to keep the satellite attitude and to optimize the total operation torque of the system-is developed. By means of this control method, an interesting torque optimum behavior is observed and a practical target capture operation is exhibited by computer simulation.>

Presents a coordinated control of multiple manipulators in space free-floating robots. The authors firstly develop the generalized Jacobian matrix and equation of motion for a space robot with multiple arms, then propose a new control method optimising the sum of squared control torque in the sense of local (instantaneous) minimization by means of redundancy. The method is applied to realistic models installing a mission arm and reaction wheels, and also a redundant arm. Through the simulation study, this paper shows that the installation and utilization of the redundant arm has great effectiveness in terms of reducing the burden of reaction wheels for satellite attitude control.>

The authors establish a control method for space manipulators taking dynamical interaction between the manipulator arm and the base satellite into account. The kinematics of free-flying multibody systems is investigated by introducing the momentum conservation law into the formulation and a novel Jacobian matrix in generalized form for space robotic arms is derived. The authors develop a control method for space manipulators based on the resolved motion control concept. The proposed method is widely applicable in solving not only free-flying manipulation problems but also attitude-control problems. The validity of the method is demonstrated by computer simulations with a realistic model of a robot satellite.>