We have developed the person following robot "ApriAttenda(TM)". This robot can accompany a person using vision based target detection and avoid obstacles with ultrasonic sensors. The robot first identifies an individual with its image processing system by detecting a person's region and recognizing the registered color and texture of his/her clothes. Our newly developed algorithm allows the robot to extract a particular individual from a cluttered background, and to find and reconnect with the person if it loses visual contact. Tracking people with vision was realized by systematizing visual and motion control with a robust algorithm that utilizes various characteristics of the image data. ApriAttenda(TM) has been exhibited at Aichi EXPO 2005, and its robust functions and smooth person following capability were successfully demonstrated.

We have developed the person following robot "ApriAttendatrade". This robot can accompany a person using vision based target detection and avoid obstacles with ultrasonic sensors while following the person. The robot first identifies an individual with its image processing system by detecting a person's region and recognizing the registered color and texture of his/her clothes. Usually, the person following robot has to detect and recognize the specified person and calculate his/her position in a complicated real-life environment of fixed objects and moving people. Our newly developed algorithm allows the robot to extract a particular individual from a cluttered background, and to find and reconnect with the person if it loses visual contact. Tracking people with stereo vision was realized by systematizing visual and motion control with a robust algorithm that utilizes various characteristics of the image data. The developed algorithm uses several analyses to extract information on the distance to each feature point, speed of target, color and texture of clothes for a stable tracking in many situations including changes of view due to self motion, shifts in lighting and objects similar to the target. The person following robot ApriAttendatrade has been exhibited at Aichi EXPO 2005, and its robust functions and smooth person following capability were successfully demonstrated

This paper describes a smart recognition system which performs character matching by replacing speech parameters with the five Japanese vowels and a few consonant categories. The proposed algorithm can make speaker-independent voice recognition. The algorithm has an advantage over the conventional speaker-independent word recognition system because it can reduce the required memory to about 0.5% of the conventional algorithm for storing the reference templates and for the instruction set, and can be performed even in a low-speed processor. We implemented this recognition algorithm in a fixed-point, 20 MIPS digital signal processor board with a 9-k/spl times/16-bit on-chip RAM. Recognition experiments using 20 Japanese city names had a 90.3% accuracy. Such an accuracy is good enough for a voice control system.

We developed a robot system which can play one-on-one volleyball games with a human player. It can not only hit a flying ball back to the player correctly but also have various autonomous functionalities to interact with humans necessary for playing games, i.e., it can pick up a ball of a requested color by recognizing voice instructions, can locate the players by recognizing their faces, and can shake hands. This robot performed the entire game sequence over 100 times at a company-sponsored exhibition, where the ball rally continued over 25 times with an arbitrary player. In this paper, we mainly describe the real-time visual feedback system of the robot which is essential for playing ball games.

As an initial stage in the realization of an autonomous robot system for the construction of overhead distribution lines, we have developed an experimental setup with a task planner to install a power switchgear model on a mock-up utility pole. When an operator inputs a task level command of a standard installation task to the system, the task planner generates robot control commands to perform the desired task automatically by referring to the environment and the task knowledge database. The robot then moves and executes the specified task according to the control commands that have been generated. We have verified in the experimental setup that, using the developed task planner a number of typical task for installation of the power switchgear model can be performed smoothly with a small amount of assistance from the operator.

For part III, see ibid., pp. 1237-1244. A simple stiffness control method for a seven-degree-of-freedom redundant robot arm is presented. The method is very practical method for application to a real robot system. The robot arm is able to move compliantly by this method without heavy computation. The method uses the transpose of the Jacobian matrix to transform the end-point forces into the joint torques for force feedback, but does not require either inverse of the Jacobian matrix or solution of the kinematic equation. Also, the method uses the "joints potential" but does not require pseudo-inverse of the Jacobian matrix to keep the configuration of an arm even if the arm is redundant. The authors have applied this control method to their dual experimental robot arms. Both arms are anthropomorphic and have seven degrees of freedom except for the one degree of freedom of the gripper and are part of the model-based remote maintenance robot system TAROS (Toshiba Advanced Robot System). The arms were able to carry out compliantly such tasks as turning a crank or fastening a bolt and to grope on the surface of the work object and locate the positions of holes.

For part I, see ibid., pp.1225-1230. The authors are developing a model-based remote maintenance robot system which autonomously carried out maintenance tasks in nuclear facilities, for example nuclear power plants and nuclear fusion reactors. Two measurements methods for the robot system are described: one is a laser range finder, and the other is a groping method for the target tasks of inserting a bolt and setting a nut. In the groping method, an arm with a bolt gropes on the flange surface, and a rapid change in the force signal is detected when the bolt tip comes across a hole. Position measurements are made by the joint angles in groping. Therefore, it has the same merit as direct teaching in industrial robots. The robot system has carried out inserting a bolt and setting a nut on a flange by using these methods.

Describes the outline of a model-based remote maintenance robot system for nuclear facilities such as nuclear power plants and nuclear fusion reactors. This robot system aims at autonomously carrying out maintenance tasks which the operator directs by task-level commands, for example, "set flange" and "set valve". This robot system is composed of a device for measuring the position of the object to be handled, a task planner which generates the robot control program from the task-level command, and an arm controller which moves the arm under either position or stiffness control according to the generated program. In experiments, the robot system carried out the insertion of a bolt and the setting of a nut on a flange.

For part II, see ibid., pp. 1231-1236. Introduces a task planner for the model-based remote maintenance robot system called TAROS (Toshiba Advanced Robot System). TAROS's task planner has a task knowledge database describing procedures for performing the desired tasks and an environmental knowledge database describing geometrical characteristics and arrangement data for the objects to be handled. These data are represented by frames. When the operator inputs a task command, the task planner generates robot control programs automatically by referring to the databases. In the task planning, even though object's position or orientation data in the environmental knowledge database include errors, suitable motions are selected from the task knowledge database according to the number of errors, and executed. Thus a desired task can be performed easily without making a detailed robot control program in which the object's position and orientation errors are taken into consideration. It was verified that the fastening and loosening operations for a nut and a flange bolt were performed autonomously by using the task planner of TAROS.

Syllable-based F/sub 0/ units (SBUs) are proposed for F/sub 0/ contour generation, along with a two-stage strategy. The two-stage strategy provides a flexible F/sub 0/ generation framework by introducing a global model and local model. The local model consists of the SBUs which make it possible to estimate the F/sub 0/ contour precisely using segmental information. Experimental results show that the proposed approach can generate a good global model (the measured multiple correlation coefficient is 0.843) and can precisely estimate average F/sub 0/ (the measured multiple correlation coefficient is 0.875). It is also confirmed that generating SBUs according to syllable positions is important in precisely estimating the F/sub 0/ contour. Listening tests show that speech synthesized with the proposed model is preferred to the output of the conventional model. It is expected that the approach will prove to be useful and powerful for synthesizing various types of speech.>

A new word identification method is proposed to determine text readings and their accentuation for Japanese text-to-speech conversion system. Since Japanese writing characteristics results in an enormous number of word candidates, a technique of reducing the number of identification trials is introduced employing a dynamic programming method. Moreover, plausibility evaluation functions are utilized in the identification algorithm to reflect Japanese writing characteristics : grammatical word transition tendency, phrase number minimization principle, longest matching advantage and unusual kana(Japanese syllabary) expression penalty. Applying this method to 299 Japanese sentences consisting of 6867 words, 95.2% of the words were correctly identified. Further precise analysis confirms that about two-thirds of the errors can be easily corrected by revising the word transition networks and by enriching the word dictionaries. The remaining errors however can be recovered only by fine semantic analysis. Through this experiment, the proposed word identification method is shown to be useful in a Japanese text-to-speech conversion system.