Human arm postures were clasified into two categories : one degree-of-redundancy posture and two-degrees-of-redundancy posture. In each case of two categories, hypotheses for determining anthropomorphic manipulator postures, which are like those of human arm, were proposed. Validity of the proposed hypotheses was verified by simulation with a seven-degrees-of freedom anthropomorphic manipulator.
When a hammer is used by a robot, it is driven by a rotational actuator such as a motor in many cases. If the length of the hammer head is not negligible compared to the link length, the hammer generates a tangential velocity to the face of an object at a hitting. This velocity component disturbs to drive a nail perpendicularly, for instance. This paper shows that a robot with a flexible link hammer can hit an object from a normal direction with only a specified normal velocity. In order to lead the hitting condition, we obtain the velocity vector of the tip of the hammer driven by a motor. And we determine the hitting time and a posture of an object to be set so that the hammer can flap it and make the tangential velocity zero at a hitting. Nailings are experimented by a 5 degrees of freedom manipulator with a flexible link hammer. It is shown from the results of the experiments that the robot can drive a nail into a wood perpendicularly on those hitting conditions.
A new three-degree-of-freedom parallel mechanism called PMP (Parallel Moving Platform) is devised to develop a smart end effector which adds dexterous capability to a long space manipulator. The PMP consists of a platform connected to a fixed base by three linear actuators and three parallel constraint mechanisms, and it has the advantages of compactness, simple kinematic relation, and translational mobility in the direction of approach. The kinematics are expressed by simple algebraic formulations. This paper deals with the kinematics analysis of the PMP and its application to the smart end effector. The forward and inverse transformation, singular configuration, and work space analysis of the mechanism are described. Moreover, the development of the smart end effector using the combination of the PMP and a 3-axis gimbal, and experiment results are presented.
This paper describes a method of identifying a target object by two mobile robots. Assuming that each mobile robot has a local coordinate system, one mobile robot detects the location of the other robot which is perceiving a target object and extracts its line of sight from the configuration of the mounted TV camera. Location of the target object is then recovered taking these results into account. Performed experiments were satisfactory.
This paper deals with a robotic system which is required to exhibit adaptability and have the capability of self-organization, and self-evolution. One such robotic system to realize this type of configuration has been studied by the authors. It is called a cellular robotic system (hereinafter call CEBOT) . CEBOT consists of many kinds of functional units. Each unit is called a “cell”, which possesses at least one function, and its own intelligence. The fundamental concept of CEBOT is based on biological systems which are adaptable and capable of self-organization, and self-evolution. Like the biological system, many cells can organize a structure to integrate the tiny functions of single cell. Therefore, the whole structure can execute complicated tasks which can not be carried out by a cell. This paper covers essential issues to develop a type of multi-agent robotic system as follows : the fundamental concept of CEBOT and the definition of the “cell”, the control law for cells to realize the “function amplification” which is a newly proposed concept, the hardware and the software configuration, and the optimizing method of the structure for both the hardware and the software of CEBOT.
Robots are expected to expand their usages into more complicated tasks and in unstructured environments. As an example of such a task, peg-in-hole task by robot is studied in this paper. We report on a computer simulation and an experiment of peg-in-hole task with no edge chamfers, executed by a robot with a force sensor mounted in the wrist applying an appropriate insertion procedure and a hybrid (position/force) control. The robot is a planar Cartesian robot with three degrees of freedom (translation with two degrees of freedom, rotation with one degree of freedom) . A condition for the peg to be jammed in the hole under the force control mode is also studied. When the robot insert the peg into the hole, stronger vibrational behaviors both in the contact force and position signals appeared. Therefore we studied vibrational characteristics to find a method to reduce the vibrations.
Dextrous manipulation of a non-articulated multi-fingered hand is presented. To achieve dextrous manipulation with a non-articulated multi-fingered hand, a parallel-jaw gripper which has a turn table at each finger tip has been developed. The turn tables in this hand are controlled by ultrasonic motors, and the rotating angle is detected by an encoder located under each turn table. This hand has only one free rotation at each finger tip, but can achieve various manipulations. For example, regrasping an object about the center of the turn table, rotating a nut, sliding a block along a plane surface, pulling out a long stick from a cylinder, etc, are all feasible. The design of this hand and manipulation using this hand are more easily realized than that of an articulated multi-fingered hand. In this paper, the mechanism of the new hand is presented, and the concrete manipulations that can be done with this hand when an object is placed in the lower pair is analyzed. The required conditions for achieving these manipulations are calculated. Since these manipulations can be achieved without moving the arm, this means the hand can effectively manipulate objects in very narrow workspaces.
Recently, a lot of research on master-slave manipulators with different configurations (MSM - DC) has been conducted, but the meneuverability of such systems has not been discussed enough. Thus, an MSM-DC using an orthogonal type master arm has been developed and experiments are performed to evaluate the maneuverability of the MSM-DC. The results are as follows : The MSM-DC can be operated as easily as a conventional MSM using a replica master arm. The slave arm position and the reflected force at the master arm are correctly generated using coordinate transformations between the slave and the master arms. The master-slave mode is suitable to perform a fine task compared with the joystick mode. The function of changing the transformation point, which is newly proposed, from the master arm to the slave arm is effective to carry out a task which has a revolving point. Additionally, quantitative evaluation for the maneuverability of an MSM is investigated using a pattern trace.
In order to develop “Active Human Interface (AHI) ” that realizes heart-to-heart communication between intelligent machine and human being, we've been undertaking the investigation of the method for improving the sensitivity or “KANSEI” communication between intelligent machine and human being. This paper deals with the mechanical aspects of “Face Robot” that produces facial expressions in order to express the artificial emotions as similar as human being. As actuators move the face skin of the face robot, we select the flexible microactuator (FMA) driven by air pressure for the sake of moving the control points of the face robot corresponding to Action Unit (AU) s, and we design the aluminium frame of the Face Robot for disposing the FMA and then we make the skull frame onto the aluminium frame. The skin of the face robot made by silicon rubber covers up the skull frame. Then we undertake psychological recognition test by showing the face images of 6 basic facial expressions expressed on the Face Robot, and the results show us that the correct recognition ratio accomplishes 83.3% for 6 basic facial expressions. This high ratio shows the possibility of Face Robot as a “KANSEI” communication medium between intelligent machine and human being.
Computed torque control is one of the control methods to drive a robot along the desired trajectory as closely as possible. However, this method has been rarely applied to industrial robot. One reason is that the calculation of joint driving torque is complicated and time-consuming. Another significant reason is that the control input based on the dynamical model with rigid links does not well compensate the movement of actual industrial robot which has transmission mechanism of low stiffness. This paper proposes a modified computed torque control method that can be applied to robots with such transmission mechanism, and a class of desired trajectories to realize this control. This class of desired trajectories is also available for trajectory generation of robot arm.