This paper describes an efficient design method for dynamic compliance control by taking the advan-tages of direct drive. First we analyze dynamic bchavior of compliance control system having force/torque sensors attached to various locations such as wrists, arm joints and actuators. Pole-zero locations and natural frequencies as well as stability issues are compared for each control system. It is shown that the dynamic performance depends on the strucural stiffness and mass distribution as well as relative locations of actuators and sensors. An efficient structure of mechanical construction and actuator-sensor locations is then drived. An optical torque sensor built in a direct-drive motor is developed on the basis of the analysis. The compliance control system equipped with the torque sensor is synthesized in such a way that an arbitrary compliance be accommodated while placing all the poles at arbitrary locations. It is shown through experiments that the compliance of the torque control system can be varied from a small value to an extremely large value without sacrifice of the system stability.

View full abstract

A prototype model of a multifunctional robot vehicle “MRV-3” has been developed. The vehicle has three locomotion modes of wheel, crawler and leg. In the wheel mode, the vehicle can go forward or reverse, sidle and rotate on a spot. In the crawler mode, the vehicle can climb up stairs, move on a rough terrain, and cross over ditches. In the leg mode, the vehicle can stand up and stride over obstacle. These locomotion modes are realized by a drive mechanism which consists of four motors and 12 electro-magnetic clutches. The weight of the MRV-3 is about 150 kg including batteries, and the dimension is 770 W × 650 L × 810 Hmm in the wheel mode. In this paper, we discuss the sequence of motion for striding an obstacle such as a pipe, and analyze the static stability of this motion and the geometric size of obstacle which the vehicle can stride over. The results of experiments based on these analysis are also reported.

View full abstract

To provide multi-degrees of freedom (DOF) for the robot system is essential to enhance its versatility. However, the output power and weight ratio of the conventional actuators are limited. The thoughtless introduction of the multi-DOF to the robot greatly increases its weight. The paper proposes a new design method, named “coupled drive”, which minimize the weight of the multi-DOF robot system. The coupled drive is to couple the multiple actuators of the robot system as much as possible in the normal driving mode and let them produces the power cooperatedly. From the standpoint of controllability, decoupling of the actuator system has been recommended. The paper proposes opposite design method from the hardware point of view. As to the evaluation function of coupled drive, an actuation index η_c is introduced. It is the ratio of the power demanded for the task to the total maximum power of the actuators. To test the validity of the proposing design method, a computer simulation is performed. A wall climbing robot, which walks on the wall with four legs by sticking the wall with vacuum sucker or magnet of the sole is used for the simulation. The optimum mechanism and walking motion of the robot is calculated by using linear programming. It is shown by the experiments that the introduction of the coupled drive greatly improves the weight and velocity performance of the multi-DOF robot system.

View full abstract

The Magnetically Supported Intelligent Hand (MSIH) is a unique device for automatic operations of precise insertion, used as an end-eff ector of robots. This paper explains its structure, control system and functions, and describes its experimental system and the experiments of automated assembly by the MSIH. The MSIH has a structure similar to that of the totally active DC-type magnetic bearing system. In the MSIH, the hand unit, gripping a work piece, is supported without any mechanical contact at the wrist by the attractive forces of electromagnets feedback-controlled with out-puts of the gap-sensors. That structure enables the MSIH to have three major functions : 1) the variable compliance function (the adaptable RCC function) ; 2) the precision-actuating function of five degrees of freedom of a work piece ; and 3) the sensing function of insertion forces/moments and displacements of the work piece. The MSIH can automatically achieve precise insertion operations in various levels and conditions by selecting or combining these functions. Especially it can have the same function as the Remote Center Compliance (RCC) . Further it is expected to break through functional limitations of mechanical elastic hands such as the RCC. The experimental system has been made to demonstrate the functions and utility of the MSIH for automated assembly and to analize insertion process. The system has a digital controller with a digital signal processor. The MSIH has been successful in very precise insertion operations in the experimental system.

View full abstract

This paper is concerned with a coordinative manipulation problem by a couple of robot manipulators in the two dimensional space. In order to manipulate an object accurately by manipulators, we must take account of not only the dynamics of manipulator but also the dynamics of object motion. Therefore, first, taking both dynamics into account, we derive a basic equation for coordinative manipulation control with two manipulators in the two dimensional space. In particular, we adopt a simple vector for the expression of object orientation instead of usual expression such as Euler angle, and Roll-Pith-Yaw. Second, based on this equation, we design a control system which achieves the desired position and orientation of object independently, by applying an existing control method for constrained dynamics. Finally, we give a simulation result to show the validity of our method.

View full abstract

Three dimensional shape measurement is very important subject for recognizing the shape of three dimensional objects. Moiré topography is very usefull method for three dimensional shape measuring method, for the moiré image is formed as a contour line image and includes three dimensional shape information. On practical use, this moire topography has several problems ; formation of high-contrast and uniform moiré image in light room, distinction method between depression and elevation, and simple extraction method of three dimensional shape information by image processing using computer. This paper presents a new type moiré formation method by using slit laser scanning and image accumulation of TV camera, simple distinction method between depression and elevation, and automatic measuring system by image processing. This new type moire formation method enables formation of moiré image in light room that is real industrial environment. Furthermore, proposed simple distinction method between depression and elevation is based on binary image distincted whether the gradient of measureing surface is positive or negative. This binary image is formed by signal arranging the TV signal of deformed grating image at the same time the moire image is formed.

View full abstract

This note is concerned witha new approach to design an adaptive control law for robot manipulators with elastic joints. Our method based on the Lyapunov's second method. This control law uses angular acceleration of rotors, but we need no information of joint elasticity. And we can apply our method to robot manipulators with even nonlinear elastic joints.

View full abstract

A model reference adaptive control based on Popov's hyperstability theory is applied to trajectory tracking control of a two link SCARA type manipulator, and its effectiveness is examined experimenttally. A linearized state equation with an optimal state feedback is used as a reference model, and such structural resemblance of the model and the real system is useful to simplify the design of the adaptive control system. Further, by introducing a new adaptation law which consisits of a combination of linear and nonlinear control, it becomes possible to reduce high frequency chatter in the control input. Results of simulation and experiment show that, by adding a c comparatively simple adaptation mechanism to a usual state feedback, the effect of Coulomb friction and that of parameter identification errors are removed and the accuracy of trajectory tracking is highly improved.

View full abstract

In the next generation model based robot systems, absolute positioning accuracy of a robot is becoming more important to effectively utilize advanced off line teaching and programming techniques instead of repeatable relative accuracy in the conventional teaching-playback methods. This paper describes kinematic modeling and calibartion of a direct drive manipulator to improve absolute positioning accuracy of a robot and an integrated robot system by estimating model's parameters and the robot's base coordinate system. Parameters in the kinematic model of a robot are classified into internal and external groups. The internal parameters are joint encoder's offset values of each axis, link lengths, displacement and twist of consecutive axes, and so on. The external parameters consist of the pose of robot's base coordinate system and are used to connect a robot with other robot systmes. The kinematic model of a robot, represented by general vector coordinate system, allows redundant parameters. Optimal set of calibration parameters is obtained by using singular value decomposition method of a Jacobian matrix. The calibration method employs the Newton method to estimate un-known parameters iteratively. Input data to this algorithm is a set of joint encoder values and 3 D position of the reference point measured by a 3 D measuring device in various pose of a robot hand. As the result of the calibration experiment, average distance error between 3 D real values and computed 3 D values by calibrated parameters is about 0.4 mm. Thus absolute positioning accuarcy of the DD robot is improved considerably. Consequently, this calibration system provides a foundation for realizing an off-line teaching and online robot programming system by establishing absolute positioning accuracy of an integrated robot system such as multi-manipulation and hand-eye coordinatin systems.

View full abstract