This paper describes a method to refine a 3-D scene description for a mobile robot selectively by attentive observation. The 3-D scene description is constructed with the constrained Delaunay triangulation and a path to a given destination is determined from the description. If the path is not found, the triangles of the Delaunay triangulation are classified into two categories: impassable and undefined. And a set of undefined triangles which is on the most promising path is observed attentively. The new data are obtained with a higher resolution. The new data, however, include the position displacement by the errors of the pan and tilt angles of the camera platform. The both images are registrated by matching segments in both images and by computing the displacement. The new data are integrated to the original description and a path is searched for in the new description. Experiments have been performed for an indoor scene which includes chair, desks, computers and so on. The results have proved that the proposed method is valid for cluttered scene.
This paper proposes a continuous hitting by a robot with a flexible link hammer. It is assumed that the hitting makes use of only the first mode in the vibration of the flexible link. The relative displacement of the first mode vibration is expressed as the function of the pulses of an angular acceleration which drives the link. This expression can easily derive two equations for the hitting conditions that the hammer can flap an object with a desired hitting velocity. The generated time and width of the pulses are determined from the equations so that a continuous hitting can be achieved with the minimum energy and hitting cycle. However, the hammer head collides with an object twice in one hitting cycle due to a larger coefficient of rebound between them. Therefore, this problem can be evaded by increasing the number of the pulses. In experiments an optimal regulator is used in order to suppress the second and the higher modes of the vibration. And a proposed continuous hitting is realized by the mode control of the regulator. As this system utilizes flexibility of the link effectively, it is more excellent than an intermittent hitting because of less energy and shorter hitting cycle.
It is very effective for developing a high torque motor to keep air gap as small as possible between a rotor and a stator. By polishing the surfaces of cylindrical rotor and stators impregnated with epoxi resin, we developed a high torque motor of a hyblid type with extremely small air gaps which has no usual bearings. The rotor and stators with smooth surface play the role as a bearing. All stator poles of VR motors are not simultaneously excited, since reversal torque is generated when rotor teeth leaves stator teeth. As only attraction force is employed, halves of stator poles don't generate torque. We investigated a uniqe hyblid type motor with rare-earth bar magnets in rotor slots, in which all stator poles generate torque by employing not only atrraction force but also repulsion force. Analizing magnetic field of the hyblid motor with MAGNA/FIM, magnetic field analising software, optimum air gap geometry were aquired. We could develop a greatly higher torque motor than a VR type motor with the same size which was developed before .
The visual servo system is a robotic servo mechanism that incorporates the vision sensor in the feedback loop. The conventional static look and move scheme is not suitable for visual servoing because the visual servoing is used mainly in dynamically chaning environments. Although the dynamic couplings exist between the robot and the camera, previously proposed schemes neglected the robot dynamics. Therefore, they are not sufficient for real-time dynamic visual servoing. This paper proposes a nonlinear controller and a nonlinear observer for the visual servo system. Also a consideration on the tracking condition is given. The nonlinear controller compensates the robot and camera characteristics. The nonlinear observer estimates the object velocity and predicts the object motion. The observer-based control scheme is proved to be asymptotically stabile. Moreover, the effectiveness of the nonlinear observer approach is verified by simulations and experiments on a two link direct drive robot.
A control scheme for grasping and manipulation by arm-hand mechanisms is proposed. The arm-hand mechanisms, discussed in this paper, consist of arm and multifingered hand, and they have the same feature as that of redundant macro-micro manipulator about manipulation. That is; 1) manipulation is possible not only by the arm but also by the hand, 2) the hand is suitable for compliant motion compared to the arm because of the small inertia, and the motion range of the arm is larger than that of the hand. The control scheme can utilize both of the merits of the arm and the hand. Simultaneously, it can realize secure grasping during manipulation. Several simulation results illustrate the validity of the proposed control scheme.
This paper describes a neural network based self-learning control system developed especially for autonomous underwater robots. Artificial neural networks can be effective tools for handling difficulties such as nonlinear dynamics and unpredictable disturbances which are often involved in the control problems of underwater robots. Inspired by growing-up processes of biological creatures, a self-learning architecture of neural net controllers has been established by the authors. The principles of the architecture called “SONCS: Self-Organizing Neural-net-Controller System” is presented and the system's feasibility is examined through several real-world applications which include constant depth and altitude swimming of a cruizing type autonomous underwater robot “PTEROA”, and vibration control of a multi-degree of freedom structure with a damping controllable dynamic damper. It is shown that the neural net controllers can be appropriately adjusted by the proposed self-learning procedures and that the SONCS can be applied to a wide variety of control problems with just easy modifications. New ideas and future perspectives, which include a newly developed quick adaptation methods called “Imaginary Training”, are discussed to make the SONCS system more attractive solution for control engineering of actual robotic systems.
A free-flying space robot with a 6-DOF manipulator cannot follow an arbitrary trajectory in the 9D generalized coordinates (3 of the satellite attitude and 6 of the manipulator joints) with only manipulator joint control, though it was shown to be commonly controllable in the literature. In this paper, we propose a method to approximate an arbitrary 9D path by introducing a perturbation around the specified path. We call thus approximated trajectory a “spiral motion”. A computational scheme for the optimal spiral motion is presented, and is followed by computer simulation. Relationship of singular points and computational convergency is also discussed.
As the first stage of the biped walking adapting to an unknown uneven surface using an anthropomorphic biped walking robot. The authors developed a biped walking robot WL-12 R VI (Waseda Leg No. 12 Refined VI) that has an ability to detect the landing surface during dynamic walking. In this paper, as the second stage of the biped walking adapting to an unknown uneven surface using an anthropomorphic biped walking robot, the authors introduce a biped walking control method for adapting to a horizontally unknown uneven surface. The authors performed walking experiments with the robot using the control method. As a result, dynamic biped walking adapting to a horizontally unknown uneven surface was realized. The maximum walking speed was 0.96[sec] per step with a 0.3[m] step length, and the adaptable step height deviation was from -12[mm] to 12[mm].
In this paper, we design a Cartesian nonholonomic robot using two balls and two actuators. It is shown that we can control not only the position of the robot hand very easily as we do for the usual commercial Cartesian robots, but also the position and orientation of the robot hand with only two inputs. To do the latter, we first transform the kinematic model of the robot into a chained form for which we propose a stabilization law. Simulation results are presented to demonstrate the effectiveness and performance of the proposed control method and a control method originally developed by Sørdalen.
This paper describes a motion planning system for intelligent manipulations using a parallel two-fingered gripper which has multiple manipulation skills. The basic approach of this planning system is the state-space approach which is described in states and operators and solves problems by searching state-space from initial state to goal state. The operators correspond to the manipulation skills and transform a state into another state. This paper proposes the methods to define the operators and the states of the environment and to search the state-space. It can plan motions for intelligent manipulations by combining the manipulation skills according to the environment. The key idea behind this approach is to augment the handling ability of a parallel two-fingered gripper, instead of developing a difficult and complicated control method that uses a sophisticated but weak multi-fingered gripper. To actually implement this idea, we have developed a new, small and simple mechanism attached to a parallel two-fingered gripper. It can rotate an object around the axis between the finger tips by the mechanism in addition to picking, placing and sliding an object.
This paper describes a method to track moving objects with multiple cooperative visual attention windows. The method is based on an implementation of transputer-based multi-processor vision system with a correlation processor. The vision system provides a large number of attention windows, each of which can be tracked using fast correlation operation performed in a special LSI MEP. On the top of the system, we built a software manager which enables to program image processing, motion control of a window, and cooperative control of windows. The cooperation is achieved using constraints, which have tree-structure, between windows. Cooperation of multiple attentions enables to track structured objects in natural scene. As examples, human arm tracking and passing car tracking are shown.