Isao Todo

Position / Force Hybrid Control of a Manipulator with a Flexible Tool Using Visual and Force Information

Robots used for industrial applications such as welding, painting and object handling have been common for many years. In recent years, the development of domestic robots has become more and more important because of the large and growing population of aged people, especially in Japan. To assist people in their daily lives, a robot must have the ability to deal with not only rigid objects but also the deformable and fragile objects usually encountered in our daily life. Many control algorithms have been developed for the manipulation of rigid objects, and in the recent past many studies related to robotic manipulation of deformable objects have also been reported (Hirai, 1998).

A Human-Safe Control for Collision Avoidance by a Redundant Robot Using Visual Information

Publisher Summary With the rapid development of service robots, avoiding collisions with human beings is a most important requirement of robot control. This chapter proposes a control method by establishing a virtual potential field around a robot to avoid collisions between a human being and the robot by the redundant joints of the robot. The greatest advantage of the proposed method is that a possible collision can be avoided automatically without requiring any action by nearby human beings. A control method is also proposed for a redundant robot to simultaneously carry out a contact task with its hand and perform collision avoidance with its redundant joints. The effectiveness of the proposed method has been demonstrated by experiments.

Control of Polishing Work by Robot for a Difficult Case of Direct Teaching (Method for Acquiring Skill from Experts in Teleoperation and Its Control Application)

Finishing is still owing to human operators because of difficulties of controlling cutting volume and tuning of the control parameters in current control methods. This paper proposes a polishing control method based on human operators teleoperation model. During teleoperated polishing work by an expert, the status of a master slave system is monitored to identify a teleoperation model of the polishing. A polishing robot control method is developed based on the teleoperation model. Such a model consists of two sub-models : one is for describing predetermined force decided by static information such as the material of the work piece, polishing tool, etc. ; the other is for adapting applying force to dynamic variation of polishing process. The effectiveness of the proposed method has been demonstrated by polishing experiments for two kinds of materials.

Kinematic Analysis of Parallel Mechanisms with a Constraint Chain of Motion

A closed loop mechanism with plural active connecting chains and one passive constraint chain, which restricts the motion of an output link, can be considered as a parallel mechanism with a constrained connecting chain. In this paper, concise kinematic analysis formular for this type of mechanisms has been derived. By modeling the passive constraint chain as an active connecting chain with some fixed active joints, equations for analyzing a parallel mechanism can be directly applied. The basis of the motion space of the constraint chain is firstly determined and that of the connecting chain is determined such that it contains all the coordinate vectors of the constraint chains. The basis of wrench space is determined as the dual basis of that of a connecting chain. The determination of the basis in this way makes it possible to separate the elements of wrenches to elements involved in the motion and other elements. This separation makes it possible to reduce the dimension of the Jacobian of the output link to that of the motion space of the constrained chain, and the motion and force analysis can be expressed by this Jacobian. Developed method is examined by appling to a 2 DOF prismatic kinematics to derive torques of active pairs verses inertial force of links.

Learning Control of a Direct-Drive Robot with a Variable Table : Position and Force Control in Constrained Work Space

In this paper, a learning control algorithm for the motion and force of a direct-drive robot, which has a variable table and six degrees of freedom (d. o. f.), is proposed for a task in constrained work space. The five-d. o. f. relative motion of an end effector attached to the wrist of the robot arm and the internal force between the end effector and the table through an object are achieved by utilizing the redundancy to avoid interference with the object. A hybrid controller for an end effector and a position controller for a robot arm are used. Under the over-constrained conditions, which are imposed by the two tasks stated above, a learning control algorithm using neural networks (NNs)is employed to control an end effector with higher priority than the robot arm. The proposed control algorithm is applied to an experiment on the control of both the position of an end effector and the contact force on the surface of the object while avoiding interference with the object. As a result, it has been shown that the trajectories of robot joints can be generated in real-time computation and a high accuracy of the end effector is obtained after the completion of learning. The effectiveness of the proposed control algorithm is demonstrated by the experiment.

Learning Control of Position and Force of a Direct-Drive Robot with a Variable Table Using Neural Networks.

In this paper, a learning control algorithm using neural networks (NNs) for controlling the motion and force of a direct-drive (DD) robot with a variable table, which has six degrees of freedom, is proposed. The motion of an end effector attached on the robot arm and the internal force between the end effector and the table through an object can be obtained by the DD robot arm in cooperation with the variable table. The control of the motion and the internal force is achieved by hybrid controllers and a feedforward controller using NNs. After the completion of learning, the NN acquires a model of inverse dynamics for the DD robot arm and the DD table. Thus the outputs of the hybrid controllers are nearly equal to zero. The effectiveness of the proposed control algorithm is demonstrated by an experiment on the control of both the position of an end effector and the contact force on the constraint surface of the object.

Application of systemic mechanisms in living beings to robot systems design

Thanks to breakthrough by science for comprehensive understanding of systems, several principles of systemic mechanisms have been revealed. Expected applications of living mechanisms through the principles are proposed to create new ideas for engineering of robot systems. Among them, restricted emergent process can provide systems with flexibility, while elaboration in selection of context for interpretation can simplify information processing i.e., intelligence. In the former process, there emerges a qualitative difference of physical structures through discrete conversion from a slight difference of input. Combining these two processes, an assembly method of blocks for diversified structures in shape has been developed, in which intelligence is simplified by utilizing for the context information accumulated in the structure being assembled. It might contribute to miniaturization of robots by saving space for intelligence, as well as basic research on principles of systemic characteristics and visualization of numeric series. Referring to vulnerability, a multiple scene survey method is proposed to find serious risks for the systems in advance.