This paper proposes a method to calculate the forward kinematics of a Stewart Plaftorm type of parallel link manipulators. The Stewart Platform consists of a base platform, a moving platform, and linear actuators connecting the both platforms. First we introduce a virtual parallel link mechanism, having the same kinematic structure as the original parallel mechanism. We assume that the moving platform of the virtual mechanism is directly driven by actuators and the limbs connecting both platforms have no actuators. A trajectory tracking control system is designed for this virtual mechanism based on the resolved motion rate control, so that the displacements of limbs are controlled by the actuators driving the moving platform. We calculate the forward kinematics by numerical simulation of the control system ; the position and orientation of the moving platform of the virtual mechanism is a solution for given displacements of limbs, when the tracking error of limbs converges to zero. The method is effective for a time varying displacements of limbs as long as the displacements are differentiable about time. The position and orientation of the moving platform is obtained without error, once the tracking error converges to zero. The simulation results illustrate the method.
In this paper, we present a sensor-based navigation method for an autonomous mobile robot in dynamically changing environment with stationary and moving obstacles. A navigation task needs to navigate the robot to a destination along a predefined path, and avoids obstacles if necessary. However, the environments is always changing, and sensory information is quite limited (actually, we have many limitations of sensory processing towards a dynamically changing environment) . We have to describe an efficient navigation manner onto the robot using simple sensory information. The first step of our approach was a use of fuzzy control. Based upon results of experiments, we propose a new navigation method for an autonomous mobile robot in a dynamically changing environment. It determines a steering angle and a velocity to send to low-level locomotion system. There are two major parts in it : one is a tracing planned path task, and the other is an avoiding obstacle task. The distribution weight between their outputs is determined with a range data and differential range data between samplings.
We have proposed a new ominidirectional image sensor COPIS (COnic Projection Image Sensor) using a conic mirror for guiding navigation of a mobile robot. The azimuth of each point in the scene appears in the image as its direction from the image center. In this paper we describe a method of estimating object's location and free space and building the environmental map for the mobile robot. Under the assumption of known free motion of the robot, locations of object's edge points around the robot can be estimated by detecting their azimuth changes in the image. After estimating the location of edge points, we detect the object surface. Both sides of edges of the object surface are contiguous in the image, and have an invariant relation with the robot motion. Against them, in case of the corridor, another edges appear between contiguous edges by moving the robot. Thus, using these relations, we can detect object surfaces, and estimate the free space rrgion are updated while the robot moving in the environment. Using this method, the robot generates an environmental map of an indoor scene while it is moving in the environment. We performed several experiments in the real indoor environment.
Actuators with three degrees of freedom used for manipulators have been developed. First, the principle of conventional motors is extended to construct motors with three degrees of freedom. Next, prototype induction and synchronous type motors with three degrees of freedom have been developed and examined. Experimental results show that the induction motor has a large slide and the rotary speed varies depending on the direction of the axis of the rotary magnetic field, while the synchronous motor rotates the same rotary speed as the rotary magnetic field. With each of the types, the rotor can move in any direction by following the rotary magnetic field.
Building large-scale space structures involves many technical problems. Elasticisty-induced vibration is one of major problems, for which many material and control oriented works have been done. In this paper, we propose new type of space structure and discuss its dynamics and control. The structure is called space multibody structure. It consists of many relatively small bodies, each of which can be considered a rigid body. The bodies are connected by passive joints and make a kinematic chain. We assume one of the bidies has an orientation control device such as control momentum wheels. This type of structure would be useful for space resource strage, for instance. A major advantage of the structure is the fact that it is free from the elasticity-induced vibration. However, the structure would change its shape due to various disturbances. An important control problem is to stabilize its shape. From dynamics point of view, it forms an underactuated mechanical system subject to nonholonomic constraints with drift terms. The dynamics is to be analized and a stabilizing control is proposed based on the result. Simulation results verify its effectiveness.
New expression of dynamic manipulability of multiple robotic mechanisms in coordinated manipulation (MRMCM) is derived. The representation of dynamic manipulability polyhedron (DMPCM), which is formed by the object acceleration realized under the feasible joint driving force of robotic mechanisms, is formulated. The way of calculating its volume is given, and this volume is proposed to be adopted as the dynamic manipulability measure (DMMCM), the index on manipulability evaluation, of such systems. The method is effective for the evaluation on systems where the application of internal force and/or the static friction constraints at grasp points must be considered. Numerical examples are given to explain the idea and show their validity. The method is expected to be useful in designing and task planning ofMRMCM.
Although most robot manipulators have many transmission gears, there have been few investigations of the torque calculation considering the power transmission system. This paper presents an expression for the torque at the motor driving the links of the manipulator with a complex transmission system including planetary gear mechanisms. Though this expression is derived by NewtonEuler method, it has a closed form. The torque expression is linear in inertial parameters formed by the combination of all dynamical parameters of main links and transmission parts. An investigation of the minimum set of dynamical parameters including those of transmission parts is made. An application for an industrial manipulator with 6 degrees-of-freedom is given.
Closed link mechanisms are useful for tasks that require high speed, high precision, and large payload capacity. It is expected that this kind of mechanisms will be used widely in future. A closed link mechanism with redundant actuators, which is called redundant actuator system, has the potential to make the maximum payload larger. This paper discusses computational schemes of the inverse and forward dynamics of closed link mechanisms with redundant actuators, which are applicable to wider cases than former method. Then we give an example of cooperative manipulation by two robot arms, and the inverse dynamics is given using the proposed computational scheme. The proposed method enables us to obtain a general solution of closed link dynamics without calculating internal forces.
A self-contained type hexapod walking robot was developed in Mechanical Engineering Laboratory. It is called MELCRAB-2. This hexapod has specific features both in the mechanical design and in the control way. This paper provides a few significant guideline to design the self-contained hexapod walking robot from the point of view about the stair climbability, turning mobility, efficient foot force delivering based on the two decoupled actuation ways, GDA (Gravitationally Decoupled Actuation) and MDA (Motion Decoupled Actuation) . An approximately straight line linkage is used in the leg mechanism of the MELCRAB-2 to satisfy both of GDA and MDA. The inverse kinematics of the approximately straight line linkage was solved for position control. Tactile sensors and proximity sensors are installed on the front side and back side of each foot to detect obstacles and walking environments. Control algorithm with many motion procedures to negotiate the stairs and walking environments was developed. MELCRAB-2 could climb up the experimental stairs with the same dimensions as the real buildings by using that algorithm. MELCRAB-2 is the real self-contained hexapod robot actuated by electric motors, and it can walk with the fast speed owing to GAD and MDA. compared with the past developed walking machines.
In the field of tele-existence and virtual reality, unconstrained measurement of human head motion is essential for HMDs (Head Mounted Display) to be interactive. Polhemus sensors developed for that purpose have deficiencies of low sampling rate and long latency. This paper proposes a new method which raises the sampling rate and cancels the latency through sensor-fusion of polhemus sensor and gyro sensor information. Validity of the method is confirmed by off-line computation of real data.