In recent years, to improve productivity of organization, the computer-supported cooperative work (CSCW) systems that support cooperative work of human have been studied and many CSCW systems have been developed. Most of these systems use computer networks and transport electronic information through the networks. But in the real world, the cooperative work consists of not only electronic information but also physical information, that can be reprenented only by physical entities. In this paper we propose human-robot interface system called RT-Michele, that utilizes mobile robots in order to transport physical entities among humans. RT-Michele provides multi-agent model that can describe the cooperative work of humans and mobile robots. With the multi-agent model of RT-Michele, both synchronous and asynchronous communication and communication both via electronic and via physical information between humans and robots can be described. And we also have developed the prototype system of RT-Michele by utilizing mobile robot Einstein that was developed in our laboratory. In the paper we describe the multi-agent model of RT-Michele, the agent description language MDL/C++, implementation issues and the office automation application which is cooperative work by human and the mobile robot.
There have been a lot of auto wind instrument playing systems, but almost all of them consist of several functions which are independent each other and they are not anthropomorphic. Therefore, since 1990 we have been developing an Anthropomorphic Flutist Robot, which is the mechanical and control model being designed similar to human organs on flute playing, in order to clarify the mechanism on one's flute playing from engineering view point. In 1990 we developed the expiration-inspiration system which models human lung in flute playing, and realized that the robot blew the flute through the first octave keeping the temperament interval by controlling air beam velocity only. In 1991 we added the playing attitude control system to this robot which can position its lips on the mouthpiece of the flute, and the robot realized to blow the flute through the first and second octave keeping the temperament interval by controlling air beam velocity and playing attitude.
This paper deals with a free-flying space robot which grapples a target satellite based on a visual feedback control scheme. As the robot body is not fixed in an inertial space, manipulator motion causes changes in the view of the visual sensor. Therefore the closed loop system for the manipulator controller involves the visual data processing which normally takes long calculation time. We construct a control system which is not explicitly affected by the calculation time and analyze stability of the system. Analysis results show that a stable control system can be constructed in spite of the long calculation time, on condition that the manipulator trajectory which is not close to singular points is settled. Moreover we developed a zero gravity motion simulator which combined a computer simulation and servo mechanisms. An experimental space robot system is constructed and experiments to grapple a target are performed on the simulator. It is confirmed through the experiments that the free-flying space robot automatically grapples the target with the visual feedback control scheme proposed.
The HEXA robot is a very fast 6-DOF parallel robot. It uses a parallel mechanism, called HEXA mechanism, which consists of a kinematic chain with five closed loops and is driven only by actuators at its base. This robot consists of six very simple elementary kinematic chains, each of which has only one active joint driven by a powerful DD motor. The moving parts of this robot is made very light at no expense of its stiffness. Therefore, this robot can move very fast and precisely. In this paper, we present development of a prototype of this robot. We choose a so-called Adept motion as a benchmark to measure its capability of fast motion. Experimental results show that the prototype robot can achieve the Adept motion in 0.465 [s/cycle] and prove that the HEXA robot is suited for very fast motion.
Control of constrained motion is very important in robotic assembly tasks and, therefore, in robot control, contact forces and states of the objects being assembled should always be taken into account. A human can control such constrained motion very skillfully. Keeping this in mind, we are developing a rule-based artificial dexterity system for assembly tasks. Before the artificial dexterity system is put into operation, we must make rules, for which we must do geometric analysis of the objects being assembled. The geometric analysis, however, is very difficult for a human. This paper presents a method for computer-aided generation of the rules using a geometric model of the objects. We apply this method to a task example: peg-in-hole task and, with rules generated by this method, execute the artificial dexterity system, having successful simulation results.
Current industrial robot manipulators are not accurate while moving fast, due to mechanical eigen-oscillations, nonlinearity, disturbance and so on. This paper proposes a robust control method for industrial robot manipulators with flexible joints, using an observer and a two-degree-of-freedom controller. The observer has two roles. One role of the observer is the estimation of the flexible joint states, to repress the eigen-oscillation. Another role is the estimation of discrepancy between the states of the mathematical model and the actual states. Our mathematical model includes the model of the joint flexibility and the nonlinearity. These estimated states are used not only for the vibration repression but also for the compensation of the nonlinearity and the disturbance. Therefore, the controller is robust for the nonlinearity, the disturbance as well as the eigen-oscillation. This robust control algorithm is implemented on digital signal processors, and is applied to actual industrial robot manipulators. Upon experiments, high accuracy and robustness have been confirmed.
Measuring relative positions among robots are essential to avoid collision and to coordinate one another. This paper proposes a signboard system and an algorithm to measure the relative positions among mobile robots in realtime. The system consists of a CCD camera and signboards with LEDs, whose 2D images are obtained by a CCD camera loaded on a different robot. The algorithm first separates the image into each signboard, and second detects the position and the orientation of the target robot. Error analysis made it possible to measure the orientation of the target robot in 360 degree with the minimum errors. The experiments with 3 mobile robots indicates that one robot can measure the other two position with the accuracy of 5% in distance within 0.4 second, in the range of 0.5 to 3 meters. The measuring speed is fast enough to calcuate the collision avoidance. The experiments verified the system is very efficient.
This paper describes an approximation technique of a pipe profile utilizing data collected by a wheel-type robot in pipe. A robot having H-shaped structure is treated to obtain the angular and linear information. First, we show that the pipe form is expressed as a combination of linear, curved, and reduced sections with three parameters such as direction, length, and curvature. Second, the locomotion of the robot is simulated to collect the angular information of the frame and wheels. Third, calculation procedure for reforming the pipe profile utilizing such information about the angular displacement between the frame and the links, rotated angles of some wheels, and sometimes the linear displacement of the frame's extension or compression is shown. Also the approximation errors caused by the differences of wheel radius, wheel base length and sampling distance of a driving wheel are visualized in graphs and evaluated by using appropriate error functions. The proposed approximation method shows the effect also in the case when the robot is of V-shaped or V2-shaped structure by taking compatible parameters of a robot strucure into consideration.
This paper aims at the analysis of the contact stability of impedance control implemented to a robot manipulator. It is important for next generation robots to realize compliant motion to environment. Impedance control is one of the compliance control methods. However, there are still contact stability problems that the control system is unstable when the robot arm implemented with impedance control contacts to a stiff environment. In this paper, two methods of impedance control, Hogan's method and acceleration feedback method, are analyzed on the contact stability influenced with the locations of sensor and actuator. The target impedance, and modeling error are also considered. An one d.o.f. arm grasping a fixed stiff object is modeled and simulated under the consideration of unmodeled dynamics of the arm and the elasticity of the contact surface of the arm and its object. The analysis draws the results that the collocation of sensor and actuator does not always stabilize the contact task in the case of the existence of the unmodeled arm dynamics, and the stable regions of target impedance are different on these methods.
In this paper programming tools for developing a sensor based behavior of a mobile robot are discussed. A method to describe a program is important base for designing programming tools. The authors have introduced an action mode representation for analyzing a mobile robot's behavior and a programming language ROBOL/0 to describe programs. The authors designed the programming tools based on the action mode representation and ROBOL/0, which is considered with (1) Action mode representation converter, (2) Simulator, (3) Action mode transition monitor, and (4) Action mode transition tracer. These tools have been implemented on UNIX workstation and used for programming of authours' mobile robot. And, it was verified that they are useful as a programming environment of autonomous behavior.