The creation of advanced dexterous mechanical manipulators and actuators is increasingly important for development of intelligent robots, which are capable of performing any human arm task. This paper deals with a spherical ultrasonic motor which is an application of an ultrasonic motor to such a dexterous actuator which has multi-degree of freedom. As a prototypal spherical ultrasonic motor, we have developed successfully two kinds of spherical motor with two-degree of freedom which has a spherical body driven by two pairs of ultrasonic actuator. The spherical ultrasonic motors show good controlability and high accurate positioning.
Capturing a target by a space robot in space inevitably causes impact forces on the contact points. This paper focuses on such a short contact period and proposes a new approach to model the contact dynamics. It derives equations to estimate the impulse of the contact forces considering the effects of the joint stiffness represented by the joint servo. Using the derived equations, it is shown that the impulse of the contact forces increases with the joint stiffness regardless of its posture. Numerical simulations and hardware experiments are given to validate the proposed approach.
A new design of holonomic omnidirectional vehicle and a laser guidance technique for the vehicles are presented. The holonomic mechanism allows the vehicle to maneuver in an arbitrary direction from an arbitrary configuration on a plane. This significantly simplifies control problems and improves positioning accuracy. A fundamental method of obtaining omnidirectional motion with holonomic constraints on the floor, using a mechanism with spherical tires, is described. Each vehicle can be viewed as an XYθ positioning table with an infinite workspace. For accurate guidance and positioning, a laser guidance technique is developed. A prototype system is designed, built and tested. Upon experiments, omnidirectional motion of the vehicle and the effectiveness of the guidance method were confirmed.
A multi-link system is defined as any finite number of rigid bodies interconnected by pairs with various constraints. Typical examples are the mechanisms in machines and limbs of humans. Recently, there is a need for parallel-mechanisms which have the advantages of maneuverability, lightness and stiffness. Many studies on them have been performed. Also, there are many studies on the motion control and motion analysis of living mechanisms. This paper describes an analytical approach to the kinematics and dynamics of a multi-link system using motor algebra. The redundant degrees of freedom arising from the analysis on the velocity motor, acceleration motor and force-moment motor are discussed. Throughout this paper, all the solutions of analysis are obtained by the linear calculation algorithms, such as QR decomposition, Gaussian elimination and backsubstitution. The data structure and the algorithms are based on linked lists.
This paper proposes a visual feedback control scheme for an eye-in-hand robot. Previous visual feedback schemes treated the robot as an ideal device to generate the desired motion of the robot hand and hand-mounted camera. However, they are not suffitient for high speed tasks because the robot dynamics is neglected. This paper proposes a nonlinear model-based scheme which compensates the robot dynamics. Asymptotic stability of the proposed scheme is proved. Also the effectiveness of the nonlinear approach is verified by simulations and experiments on a two link direct drive robot.
The authors have developed the finger-shaped tactile sensor which consists of an optical waveguide, an elastic cover, eight LEDs to provide light to the waveguide, a PSD for obtaining current outputs from a light spot, and lens system for forcussing the light emitted from the contact point. The use of a PSD enables us to execute a real time sensory feedback control using the finger-shaped tactile sensor. This is an essential advantage to use a PSD compared with a CCD. However, since a PSD can only provide four current components depending on both the intensity of the total light emitted and the location of the light spot focused on the PSD, it can not decompose multiple contacts into individual contact points. Assuming that multiple contacts are allowed, we challenge the issue on how to estimate individual contact points from the PSD outputs. In this paper we consider this issue under the assumtion that multiple contacts never happen at the same time. We show that the change of contact phase from single contact to double contact can be evaluated by differentiating the PSD output signals and the multiple contact points can be estimated by considering the time history of the output signals.
This paper describes dynamic analysis of a tactile sensor using suspension shell to sense not only normal force but also arbitrary force in direction and position. Notice that the angular displacements of β, γ, α and linear displacements ofρ, λ expressing the behavior of the suspension shell are given, and that the number of external forces contacting the shell is supposed to be one. Then, algebraic formulation is made to determine the magnitude, direction and position of the force uniquely by considering a dynamic equilibrium condition. Only the longitudinal force of the springs suspending the shell coaxially around the rigid finger body is utilized. Lateral force of the spring does not appear since the ends of the springs are connected by ball joints. Thus the dynamic analysis becomes simple. Calculation procedures to extract magnitude, direction and position of the external force are shown. Also, simulation to verify the proposed procedures are made. The suspension shell having similar forms of the finger body is rigid, and thus it gives the sensor robustness.
This paper presents a new control strategy of the artificial potential field approach to a real-time motion planning problem in a known environment. In the artificial potential approach, the goal is represented by an attractive artificial potential and the obstacles are represented by a repulsive one, so that a robot reaches the goal without colliding with obstacles by using a gradient technique such as the steepest descent method. Although this approach is quite simple and computationally much less expensive than other methods based on the global information on the task space, so far few attentions have been paid to dynamic behavior of the generated trajectories such as movement time from the initial position to the goal and the velocity profile of the end-effector motion. In the present paper, we argue that the dynamic behavior of the arm trajectory to be generated should be taken into account within the framework of the artificial potential field approach, and introduce a time base generator that acts as a time varying gain and determines a dynamic behavior of the arm motion. By synchronizing a potential function used in the artificial potential field with the time base generator, convergence time of the generated arm trajectory can be adjusted through the time base generator without any change of the potential function itself.
In this paper, the nonlinear three-layer Neural Network using a dynamic back-propagation method which based on the output signals of the hybrid controller and past output information of the plant is proposed for the Cartesian position/force control of multi-DOF robots. The generalized delta law is used as a learning law to adjust the weights of the Neural Network at every sampling time in order to adapt to an unknown environment. Experiments were done with a 3-DOF Direct-Drive Planar Robot Manipulator. The proposed Neural Network Controller controls the position and the angle of the end-effector as well as the applied force to an unknown environment. The experimental results show the effectiveness and flexibility of the proposed Neural Network Position/Force Controller for Multi-DOF Robots.
An ultrasonic range sensor has sophisticated characteristics whichw always depends on environmnt. So understanding of sensor characterristics is very important for programming behavior of an autonomous mobile robot. A physically based simulation helps understand characteristics of sensors. In this paper, a simulation algorithm of ultrasonic range sensor is proposed, and its implementation is dicussed. To realize fast, but accurate physical simulation, particle model of ultrasonic wave is introduced. This model includes not only specular reflection, but also diffuse reflection. The result of simulation was very similar to the result of the experiment. The simulator is fast and accurate enough for an aid of programming behavior of the robot.