A new modeling method of human behaviors is proposed in this paper. In the proposed method, it is assumed that a person changes his behavior according to the change of the situation around him, and this concept is expressed by If-Then-Rules, which are called behavior rules. In behavior rules, a human behavior is described as a discrete event, and the change of the situation around a person is described by Hidden Markov Model (HMM) which models multi-dimensional time series sensing data. Moreover, and early estimation method of the next human behavior and the timing of its execution is proposed based on the proposed human behavior model. In this research, human operations of a radio controlled vehicle are modeled as an example of application of the proposed model. The usefulness of the proposed method is examined through experimental results of behavior estimation with the constructed behavior model.

We have proposed a modeling and recognition method of human behaviors in this research. In the proposed modeling method, we have assumed that a person changes his behavior according to the change of the situation around him, and this concept is expressed by If-Then-Rules, which are called behavior rules. In behavior rules, the change of the situation around a person is described by Hidden Markov Model(HMM) which models multi-dimensional time series sensing data. In this research, a support system for human driving behaviors has been assumed as an example of application of the proposed model. In order to realize the system which shows a driver his next behavior, we propose an estimation method of the next human behavior and the timing of its execution. The usefulness of the proposed method is examined through experimental results of behavior recognition with the constructed system.

We propose a modeling and recognition method of human behaviors in this paper in order to realize such intelligent systems that can adapt humans, i.e. the systems that support humans by considering human behaviors, In the proposed method, we assume that a human changes his behavior according to the change of the situation around him, and this concept is expressed by If-Then-Rules, which is called behavior rules. In behavior rules, the change of the situation around a person is described by multi-dimensional time series sensing data which is modeled with Hidden Markov Model(HMM). To recognize the change of human behaviors, the optimal If-Then-Rule is chosen based on the current human behavior and similarity to the time series data of the situation obtained by sensors. As an example of human behaviors, human driving behaviors are considered, and a recognition system of human driving behaviors is constructed. The usefulness of the proposed method is examined through some experimental results with the constructed system.

We suggest motion adaptation method against environmental changes for service robots. Several motions such as walking, dancing, demonstration and so on are described with time series patterns. These motions are optimized with the architecture of the robot and under certain surrounding environment. Under unknown environment, robots cannot accomplish their tasks. We propose motion generation system based on heuristic search with histories of internal sensor values. New motion patterns are explored under unknown environment based on self-evaluation function. Self-evaluation function is composed of difference of internal sensor values between standard environment and unknown environment. New motion patterns are generated to maximize self-evaluation function without external information, such as run length, position of robot, human observation and so on. Experimental results show that the possibility to adapt autonomously patterned motions to environmental changes.

Robot motions are designed according to the properties such as the number of joints, the range of each joint, the length of each frame and so on. Structure changes such as joint failures or frame distortion make the tasks of prepared motions unaccomplished. Identification of structure changes on multi degree-of-freedom (DOF) robots like humanoid is so difficult that the traditional approaches based on inverse kinematics would not be applied. We propose a motion adaptation method to generate new proper motions accomplishing the tasks on changed structure without identification of the structure changes. All the motions can be expressed with the time-series data of postures. Each proper posture as an element of motions for changed structure is estimated by Kriging with joint angle data of several original postures sorted by Vector Quantization (VQ) and the corresponding postures explored on changed structure by Simulated Annealing (SA). At the case the tasks couldn't be accomplished with the generated motions in designed DOF, new motions are generated in expanded DOF adding redundancy joints. Experimental results show that the generated motion by proposed method is accomplished 96.6% similarity with the original task referenced failure motion's task to 0% on changed structure by single joint locking.

This research focuses on the recognition of vehicle external objects using selective visual attention system. We had proposed the selective visual attention system with the feedback of recognition results. The system has three visual features such as still image feature, motion feature and blinking feature, and has intention maps for each visual feature. The intention maps have come to express existing regions of objects according to recognition feedback. Motion feature and blinking feature have strong correlation with each other. That causes a deviation of feature extraction and of intention map update. Consequently, the regions of attractive objects disappear on intention maps. Blinking feature must be independent from motion feature. We propose the blinking brand-new feature extracted by FFT applied to dozens of frames. As the results of experiments, the proposed feature can affect goodly to visual attention system.

This paper presents a new framework for traffic flow control based on an integrated model description by means of a hybrid dynamical system. The geometrical information on the traffic network is characterized by a hybrid Petri net (HPN). Then, the algebraic behavior of the traffic flow is transformed into a mixed logical dynamical system (MLDS) form to introduce an optimization technique. These expressions involve both a continuous evolution of the traffic flow and an event-driven behavior of the traffic light. The HPN allows us to easily formulate the problem for a complicated and large-scale traffic network due to its graphical understanding. The MLDS enables us to optimize the control policy for a traffic light by means of its algebraic manipulability and use of the model predictive control framework. Since the behavior represented by the HPN can be directly transformed into the corresponding MLDS form, the seamless incorporation of two different modeling schemes provides a systematic design scenario for traffic flow control.

The focus of this paper is mental tension detection in speech to assist control the tension in day-to-day business such as conferences and operations in a call center. It is difficult to use classical techniques for mental tension detection in dayto- day business because those techniques require invasion body by electrodes or squirts and tied up by cables. In order to achieve a non-invasive, non-contact and low-restricting method, this proposed technique uses acoustic features in the speech. The technique uses the vocal tract model which represents the shape and the tightness of throat muscle. The Gaussian Mixture Model (GMM) classifies two mental tension states: high-tension and non-tension. The experiment result shows high recognition rate of mental tension detection.

This paper proposes adaptive emotion recognition system in speech by feature selection based on KL-divergence. In order that the system can choose the most suitable feature set for emotion recognition, we propose an evaluation method for the set of prosodic features based on Kullback-Leibler divergence (KL-divergence). Additionally, we propose a method of feature selection system, using genetic algorithm (GA) making use of rapid evaluation based on KL-divergence. Experimental results show the proposed system can acquire efficient the prosodic feature set for emotion recognition in short order without constructing a recognition system. Furthermore, the accuracy of emotion recognition is significantly improved with the prosodic feature set selected by the proposed system.

Emotion expression through behavior is meaningful for communication processes between robots and humans. Automatic emotion expression generation is needed for those robots in order to stimulate communication with humans. Proposed technique generates emotional motions by modifying base motion patterns using the combination of the adjectival expressions. In order to generate emotional motions, emotional motion modifications should be based on rules which represent the relationship between emotion expressions and the combination of the adjectival expressions. Emotion modification rules are obtained by organoleptic tests. This paper, in particular, shows the relationship between the degree of adjectival modification and emotion expressions. Thus proposed technique generates the suitable emotional motion patterns from base motion pattern.

In this paper, we describe a motion pattern acquisition system from time-series posture data for a robot. In order to extract reusable motion patterns without the target pattern information a priori, proposed system extracts the patterns observed frequently. The extracted motion patterns often include transition motion at first and last. In order to reduce the transition part, we propose common block extraction method. The similar common block is classified to the same category. Then, the proposed system obtains the average of each cluster as representative motion patterns. Experimental results show remarkable motion patterns extracted by proposed system from time-series posture data.

This paper presents a modeling strategy of human driving behavior based on the controller switching model focusing on the driver's collision avoidance maneuver. The driving data are collected by using the three-dimensional (3-D) driving simulator based on the CAVE Automatic Virtual Environment (CAVE), which provides stereoscopic immersive virtual environment. In our modeling, the control scenario of the human driver, that is, the mapping from the driver's sensory information to the operation of the driver such as acceleration, braking, and steering, is expressed by Piecewise Polynomial (PWP) model. Since the PWP model includes both continuous behaviors given by polynomials and discrete logical conditions, it can be regarded as a class of Hybrid Dynamical System (HDS). The identification problem for the PWP model is formulated as the Mixed Integer Linear Programming (MILP) by transforming the switching conditions into binary variables. From the obtained results, it is found that the driver appropriately switches the "control law" according to the sensory information. In addition, the driving characteristics of the beginner driver and the expert driver are compared and discussed. These results enable us to capture not only the physical meaning of the driving skill but the decision-making aspect (switching conditions) in the driver's collision avoidance maneuver as well.

Pulsed neural networks based on delta-sigma modulation (DSM-PNN) suitable for digital hardware implementation are introduced in this paper. Using delta-sigma modulation, which is a method to convert a signal into a 1-bit pulse-stream, it becomes possible to realize compact hardware implementation and accurate signal processing for pulsed neural networks. Since the proposed method has high versatility, various neural networks can be easily implemented in the hardware. In this paper, two pulsed neural networks, performing PCA or ICA, are introduced and their performance evaluated.

This paper presents a new hierarchical scheduling method for a large-scale production system based on a hierarchical Petri net model, which consists of FOHPN and TPN. The automobile production system equipped with 2 stand-by lines is focused as one of a typical large-scale system and these stand-by lines are controlled by binary signal. In a high level, the FOHPN model is used to represent continuous flow in production of an entire system, and MLD description is used to control the net dynamics of FOHPN. Also in a low level, TPN is used to represent production environment of each sub-line in a decentralized manner, and MCT algorithm is applied to find a feasible semi-optimal process sequences for each sub-line.

This paper presents a method for active background subtraction in sequential images taken from a moving camera on the vehicle. The active subtraction is carried out by estimating camera motion using a gyrosensor. Applying the proposed method to contour matching, we present a fast and robust system for detecting moving pedestrians.

This paper introduces a new encryption method for a block cipher. To keep the ciphertext safe, traditional methods usually use secret keys in simple fixed logic operations. To break the cipher, cryptanalysis has been developed. They use stochastic characteristics such as differential and linear relationships between the input and output data. Encryption algorithms must be secure against these analyses. The proposed method uses the secret key to modify substitution tables called S-Boxes in the encryption algorithm. They are reconfigured on demand using FPGA. Our method is hard enough to prevent the encrypted data from analyzing the secret key using statistical cryptanalysis.

/sup H/ybrid dynamical systems (HDS), which contain both discrete logical symbol and continuous signal, are attracting great attention in the field of system control. In this paper, a new symbol based control strategy for a line following control of a two wheeled vehicle is proposed. The vehicle is supposed to have a low-resolution sensor and actuator. The control requirement, however, is specified so as to keep the vehicle as close as possible to the center of the line. The controllability and observability issues are investigated, and a concrete control policy based on the continuous state estimation is proposed. Some experimental results are shown to demonstrate the usefulness of our idea.

In this research, we discuss how to automatically generate the motion sequence of actuators for error recovery, including parallel path divergence in the programmable logic control (PLC). Since the sequential path in PLC can be expressed by a state flow, the goal of our research is to find an optimal shortest path from the initial state to the final state. In order to solve this problem, we propose a new search strategy, which consists of two search stages. In the first stage, real-time A* (RTA*) algorithm is used to find a suboptimal sequential path from the initial state to the final state. Based on the suboptimal solutions found by RTA*, the second stage tries to pick up some underlying parallel path divergences among solutions obtained in the first stage. From some simulation results, we have verified that our proposed method could search a suboptimal motion sequence including parallel path divergences with less computational amount.

This paper deals with statistical mechanical characteristics of fuzzy clustering regularized with the fuzzy entropy, especially a "phase transition" phenomena. We derive the critical temperature at which the phase transition occurs in the fuzzy clustering. Then, by combining the cluster divisions by phase transitions with an adequate division termination condition, we propose a hierarchical fuzzy clustering method which can determine a number of clusters. Numerical experiment shows that the proposed method works properly.

Presents a method for control of autonomous mobile robots to acquire fine actions based on real-time search. In the proposed method, the information criterion for the environment is defined based on the Kullback-Leibler divergence, which measures the quality of the environmental information used for the action search. The robot searches suitable actions based on the environmental information which is improved step by step in the sense of this criterion. According to this, we can harmonize the trade-off between the calculation amount and information quality, and make the search process faster. The proposed method is applied to the moving obstacles avoidance problem, and its usefulness is shown through some simulation results.

The requirement of handling deformable objects such as leather, paper and rubber is growing. Since it is very difficult to make a physical model of them, the design of a controller to manipulate them becomes one of the significant problems in the field of robotics. If we look at the operation of a human worker, the deformable objects seem to be handled naturally and smoothly. The paper presents a method to design a controller for assembly tasks which involves the manipulation of deformable objects. The proposed method extracts the dynamics that human workers used in executing the demonstrated task, and embeds that in an event driven hybrid controller. In our control system, an event observer estimates the change of task state based on force and visual information like human workers, and switches the dynamics appropriately according to the task state. The proposed method is applied to a hose insertion task by implementing it in an industrial robot controller.

Presents a scheduling method for a manufacturing system based on a timed Petri net model and a reactive fast search algorithm. The following two typical problems are addressed in the paper. (1) Minimize the maximum completion time. (2) Minimize the total deadline over-time. As for problem (1), a search algorithm which combines RTA* and a rule-based supervisor is proposed. Since both RTA* and the rule-based supervisor can be executed in a reactive manner, machines and AGVs allocations can be scheduled reactively, and simultaneously. As for problem (2), the original Petri net model is converted to its reverse model and the algorithm developed in problem (1) is applied with regard to the due time as a starting time in the reverse model.

Neural network hardware is necessary for demanding real-time applications such as pattern recognition. In the paper, an improved calculation algorithm of the neuron model with sigmoid function is proposed, which is suitable for hardware implementation. The algorithm is based on the multi dimensional binary search. The neuron circuits implemented on FPGA by the proposed algorithm have shown excellence in size and circuit frequency compared with the conventional circuits with sum of product operation.

Proposes an architecture of a neuro-hardware that can be realized on a far smaller-scaled circuit compared to the conventional approach. In order to reduce the scale of the circuits, the architecture employs a method of computing the membrane potential and the sigmoidal function by encapsulating the probability properties into the relative delay between two pulses of different signal lines. The proposed architecture enables us to integrate more than one hundred of neurons on a latest FPGA chip, which means thirteen-fold miniaturization compared to conventional architecture.

A cerebellar model arithmetic computer (CMAC) is a neural network whose advantage is fast learning. It is, however, difficult to decide on the various parameters of a CMAC in advance. The pseudo-bacterial genetic algorithm (PBGA) is an evolutionary algorithm that is efficient in local searching. This paper proposes a "PBGA/CMAC" system that decides the positions of partitions, which are the main parameters of a CMAC, using the PBGA. The PBGA/CMAC hardware is implemented for acceleration, because PBGA/CMAC needs a large amount of computation time. An efficient learning method using pipelining is also presented. It is found that the accuracy with the proposed learning method is almost the same as that of the conventional CMAC's learning. The PBGA/CMAC hardware is 140 times faster than that of the equivalent PBGA software.

This paper is dealing with the fuzzy clustering method which combines the deterministic annealing approach and the fuzzy entropy. Regularizing fuzzy c-Means with the fuzzy entropy, we obtain the Fermi-Dirac type distribution function as the membership function. Then we examine the meanings of Fermi-Dirac function and the fuzzy entropy from the statistical mechanical point of view, and show that the mechanism of this fuzzy clustering is no other than the Fermi-Dirac statistics. Also we propose the algorithm which performs deterministic annealing of the membership function. Numerical experiment demonstrates that the proposed method can cluster a given set of data points properly.

We propose a new self-creating and self-organizing neural networks utilizing weight duplication. The proposed model consists of competitive and input layers. Weights of the activated node are modified by competitive learning. A new node is created depending on the activation frequency. The weights of a daughter node are duplicated from the mother node. The mother node has refractory period just after the creation. The weights of the daughter node and those of her mother node will be similar after the refractory period. A mother-daughter relationship represents the hierarchical structure of input data. It is possible to organize the column by itself with similar features which has hierarchical structure. The represented features will become precise when descending the hierarchy. Some simulations are carried out to show the feasibility of the proposed model.

This paper introduces a reconfigurable computing technique for encryption processing. Reconfigurable computing (RC) is capable of accelerating the information processing using dynamic reconfiguration of field programmable gate arrays (FPGAs). Dividing the target problems into hardware and software processing appropriately, the computation time will become much faster. A fast and flexible encryption technique is required on the Internet. An encryption technique generally consumes computational power and needs specific hardware for feasible use. Specific hardware is not capable of changing its architecture to adapt to new functionality. RC is suitable for this purpose. We implemented an RC system on an FPGA board. To examine the feasibility of this system, we apply it to encryption processing.

We propose a method based on a real-time search for the proper action of an autonomous mobile robot using the framework of the Anytime Algorithm. This is available for a real-time search in a dynamic environment. In order to apply the Anytime Algorithm to the multi-objective real-time search in the dynamic situation around the autonomous mobile robot, we adopt prediction, a switching evaluation function technique, the action watch dog and an evolution strategy. These enable the robots to realize the real time search. The paper explains the proposed method and shows the feasibility through experimental results using real robots.

A new method of determining field of visual attention using associative memory is proposed. Paying attention to the typical features of the patterns plays a great role in recognition scheme of human beings. In conventional approaches, the field of visual attention is determined by the designers a priori and fixed while the systems are running. The proposed model is capable of determining the field of visual attention automatically. The proposed model is hierarchical network which consists of associative memory layer and feature layer. The connections between the layers represent the features of the input patterns and are generated automatically. These connections are modified through the Hebbian learning rule. In the proposed model, features are extracted through bottom-up procedure. The connections corresponding to the most excited unit in the feature layer are used as attention points in the top-down procedure. Through these procedures, the system is capable of associating appropriate patterns.

Impedance control is widely used in the field of industrial world. In a certain task, it is important to decide the impedance parameters in order to realize the desired task. However, it is very difficult to calculate analytically, and the method to extract impedance parameters from human demonstration often exist unevenness in time and space in the human data. Modelling with hidden Markov model (HMM) is known as one of the promising technique to construct an efficient model for time-variant data including unevenness. HMM is capable of characterizing a doubly stochastic process with an underlying immeasurable stochastic process which can be measured through another set of stochastic processes. In this paper, we propose a method to model the series of impedance parameters identified from human teaching data with HMM as human skill model of the peg-in-hole task. In addition, realization method of the task based on the obtained model is shown.

We propose a method based on a real-time search for the optimal action of an autonomous mobile robot using an evolution strategy. By searching the optimal action for the facing situation in real time the robots need not suffer from the problem of categorization of the situations and problem of pre-definition of rules. The basic idea of the proposed real time search technique is to harmonize computational amount with a property of solution. In order to realize this requirement we adopt a two-stage evaluation technique. The proposed method also adopts the prediction and action watch dog systems. These function enable the robots to realize the real time search. The paper explains the proposed method and shows the feasibility through the experimental results using the real robot.

Since impedance control can achieve the desired dynamics between tool and environment, it is widely used for complex tasks. But deciding the impedance parameters for the task is very difficult because they cannot be calculated analytically. If we can extract impedance parameters from human demonstration, it is appropriate to use them. However, if there exist disturbances and/or noise at the playback stage, which was not taken into account at the skill acquisition stage, the executed task by the robot is far different from the human demonstration. In order to realize the robust skill, we introduce a hybrid architecture. This architecture enables robots to use the most suitable dynamics (impedance parameters) for each task situation. Moreover, because of event driven architecture, we can expect the improvement of robustness of the skill in the time domain. We apply our proposed method to the peg-in-hole task, and show some experimental results.

Presents an adaptive digital filter based on the multi-agent concept. The proposed filter consists of nonlinear small filter agents which cooperate with each other. The proposed filter is capable of adapting to a dynamically varying environment by changing the roles of the agents. At an instant, an agent is assigned to act as the main agent to minimize the error between the filter output and the teaching signal. The other agents, which assist the main agent, aim at minimizing the error of the main agent. Experimental results show good performance compared with conventional filters. The proposed filters are capable of adapting themselves to dynamically varying environments, while the conventional methods fail.

When we apply an impedance control to execute any tasks, it is very important how to decide the impedance parameters to realize the desired task. If we can extract impedance parameters from human teaching data as characteristics of the human skill, it is appropriate to use them for control because of the similarity between an impedance control and a human fingertips control. However, there often exists unevennesses in time and space in human data. Modeling with hidden Markov model (HMM) is one of the promising technique to construct an efficient model for time-variant data including unevennesses. HMM is capable of characterizing a doubly stochastic process with an underlying immeasurable stochastic process which can be measured through another set of stochastic processes. Therefore, the probabilistic modeling of certain time series data which includes unevennesses caused by the human is possible. In this paper, we propose a method to model the series of impedance parameters identified from human teaching data with HMM in order to extract an essential discrete model which expresses the human skill. In addition, some applications of the obtained model to robot control and skill evaluation are shown.

The principal aim of this study is to show how an autonomous mobile robot can acquire the optimal action to avoid moving obstacles through the interaction with the real world. In this paper, we propose a new architecture using the hierarchical fuzzy rules, fuzzy evaluation system and learning automata. By using our proposed method the robot acquires the fine behavior on how to move to the goal avoiding moving obstacles using the steering and velocity control inputs, simultaneously. Also we show the experimental results to confirm the feasibility of our method.

This paper studies the learning control problem of disassembly task sequence. The proposed learning control system has two sub systems. One of them is a supervisory control system which can realize the implicit specifications such as equipment-related or task-independent specifications. The other one is a learning automata system which can realize the explicit specifications such as task-dependent specifications. This approach allows the flexible design to accommodate frequently changing manufacturing requirements. Some simulation results to verify the effectiveness of the proposed system are shown.

Recently, a disturbance observer based controller has been widely used in the industrial world because of its simplicity. However, this method depends on the intuitive approach in determining the disturbance estimation filter, and therefore, more systematic approach is desired. In this paper, we first combine the generalized compensator and the disturbance observer based controller by clarifying the internal structure of the generalized 2-degree-of-freedom (DOF) compensator. Secondly, based on the clarified structure, we derive a robust stability condition, and propose the design algorithm of a free parameter taking into account the condition. The proposed design algorithm is easy to implement and, as a result, we obtain a lower order free parameter than that of the conventional design algorithm. Thirdly, we show by adopting an appropriate coprime factorization that the clarified structure can also be regarded as an extended version of the conventional PD compensator. Finally, we apply the proposed algorithm to a positioning control of a 3-DOF direct drive robot, and show some experimental results to verify the effectiveness of the proposed algorithm.

In the field of robotics, many researchers have studied the representation and analysis of an automatic assembly system. In previous research, however, no control-theoretic ideas for the assembly system have been found. In this paper, the authors study a closed-loop desired marking control problem for the automatic assembly system using the supervisory control technique. First, the authors represent the assembly system with a Petri net. Since the assembly Petri net has a "tree structure", the authors show that the assembly Petri net has some formal linguistic properties. Based on the properties, the authors clarify that the solvability of the desired marking control problem depends only on the controllability of a control object, and that one can construct the supervisor for the assembly Petri net with a finite automata. Finally, the authors discuss the supervisor reduction technique and give two examples.>

An automatic generation of assembly or disassembly sequence is a significant problem for an automatic assembly using robotic manipulators. This problem can be separated into following three stages. 1. Extraction of all feasible assembly/disassembly tasks. 2. Generation of all assembly/disassembly sequence. 3. Decision of optimal sequence. As for the stages 2 and 3, one of the efficient strategy based in the AND/OR graph has been proposed. However, as for 1, previous researches are not efficient. In this paper we propose a new technique for the examination of the feasibility of each assembly/disassembly task by using topological operations. The presented method is expected to execute with smaller computational amount compared with the previous methods, because a separating path of a mechanical part without collision with other parts is obtained without search of all space. A mapping, which maps shape of a subassembly to a feasible path of the other subassembly around the first subassembly, keeping contact, is introduced. An area of feasible translational motion of the second subassembly without interaction with the first is calculated. Criteria of feasibility of disassembly task are presented. Finally, we show a simple example to verify the proposed algorithm.>

An assembly planning method using Petri nets is presented. First, the assembly network is modeled using the conventional AND/OR net. Second, the state shift equation of the Petri net, which is its mathematical model, is analyzed. This analysis shows that the basis solution the of a state shift matrix is identical to the set of a solution which corresponds to an actual assembly sequence. This means that in considering an optimization problem which minimizes a summation of a weight value of all tasks included in the assembly sequence, a linear programming technique can be used. Some numerical examples are given to certify the validity of the analysis.

The transitional motion when a robotic manipulator and a workpiece first make contact is fundamental and occurs frequently. The authors propose a learning controller to improve the contact motion. The learning controller has two features. It observes input and output signals at different periods to improve the transient response of the system; and it can modify an initial state of the system, which includes an approaching velocity of the manipulator. By modifying the approach velocity, a fine contact motion can be achieved, because the magnitude of the impact force depends on the approach velocity. Simulation results verify the effectiveness of the proposed method.>

The operation of small mechanical resonator by laser irradiation is reported. The bending moment caused by the photothermal effect is advantageously used for the driving force of the vibration. The bending moment was calculated theoretically for the layered plate-like structures. The authors fabricated the micro-cantilever as a mechanical resonator by lithographic process. The characteristics of the vibration were investigated in vacuum. In addition, the flexural progressive wave was excited on the thin plate structures by using two lasers with different modulation phases.>

Most conventional trajectory planners are based on the trapezoidal pattern of the end-effector velocity on the desired path in a task space. This method can be implemented in real time. However, the motion planned by this method usually leads to a very slow motion, because it is difficult to take into consideration a constraint on the magnitude of joint velocities or joint torques. The authors propose a new trajectory planner for three DOF manipulators based on the kinetic energy of manipulators. The supplied electric power for the manipulators mainly depends on the differentiation of the kinetic energy. Therefore, the proposed method can take into consideration the constraints on the magnitude of the supplied electric power and joint velocities, i.e. one can realize a 'high speed motion' under the constraints on the supplied electric power and joint velocities. Also, the method proposed does not require much calculation and can be implemented in real time.>