An optical proximity sensor which measures a distance of an object from 0 mm to 50 mm is developed. The sensor uses a driving mechanism to cast a light repetitively to the object. A cylinder driven by the driving mechanism is located around a point source of light and is provided with a linear or curved slit. Since light radially coming from the light source passes through the rotating slit, the object is illuminated with a cycle of the cylinder's rotation repetitively. A photodiode detects a reflected light from the object. By differentiating the light signal of the photodiode, the time period when the light illuminates the surface of the object facing the front of the photodiode is obtained. The time period is utilized to calculate the distance based on a simple triangulation. The sensor has such merits that the measurement covers the range greater than 180°and is not affected by the environmental illumination, orientation and reflectivity of the object. Measurement principle and experimental results are given to show how the sensor works and the measurement is achieved without setting specific conditions.
This Paper descrives a three dimensional robot vision sensor by astigmatism beam array projector using the fiber grating. The method that was previously reported containes a problem: the measurement range is limited in order to obtain the shape of an object uniquely from a single input image. To solve the problem, a novel method that uses the two range data obtained by the method using astigmatism and the active stereo is proposed. Range data obtained by the active stereo is high accuracy. However, we cannot determine range uniquely from a single input image. On the other hand, range data obtained by the method using astigmatism is determined uniquely. However, we cannot expect high accuracy because of the laser speckle noise, quantization error of an image, distortion of CCD camera, etc. To improve the performance of the sensor, this proposed method uses the two range data obtained by these methods. In this paper, We show measurement results obtained by developed system and verify the performance of the system.
In this paper, we propose a method of planning collision-free trajectory for two manipulators along specified joint paths. The joint path of each manipulator is represented by a path parameter and two path parameters generate a two-dimensional space which contains the collision region of two manipulators. In order to connect the path parameter with time, a time scale factor is introduced. In a three-dimensionl space constituted by the two-dimensional space of the path parameters and the time scale factor, minimum time trajectories for two manipulators are searched under joint torque constraints and collision-free condition by dynamic programming. There are two advantages in this method. One is that deadlock does not occur in the search because the collision region is obtained beforehand. And the other is that the proposed method considers manipulator dynamics, therefore, not only collision avoidance but also minimization of traveling time is carried out at the same time.
The purpose of this research is to make a jumping machine jump over a big obstacle or a ditch and so on. A robot sometimes needs to jump to move over these obstacles. And the authors plan to do fundamental researches on a jumping machine to (1) make the machine jump as highly as possible, (2) control the direction of the jumping, (3) control the attitude of it in the air and (4) make it land softly. The machine is designed to jump by using actuators and springs effectively. As to the evaluation function of jumping height, a jumping height index ηJ is introduced. It is the ratio of the maximum of the shortest distance between the ground and the bottom of the machine for the jumping to the maximum of distance that the body can move along the leg. The computer simulation and the experiment are performed to prove the validity of the design method using the jumping height index. It is important to consider the machine dimension. ηJ varies with the size of the machine. It is calculated to different size of machines. It is shown by this simulation that the small machine has big ηJ and the big effect of the springs.
As one of problems that occur when multiple autonomous robots act in a common environment, task assignment is discussed in this paper. There are two patterns in terms of action of multiple-robots. Namely, one is parallel, individual action, and another is cooperative action. Object pushing problem is taken as a typical task, where both of patterns of action are necessary. At first, a method of decentralized management of situation of tasks and robots using communication is proposed. Then, algorithm for robots to process tasks related to the situation based on the managed information is developed. Finally, a simulation system for object pushing tasks and a wireless communication system between the simulators are developed. As a result of experiments using the simulation system, the tasks are processed by robots cooperatively, and the proposed method is verified. In addition, the efficiency by parallel action and the executability of high-level tasks by cooperative action are proved.
We presents a flexible and highly-reliable gray-scale vision system besed on multiple cell-feature extracting engine composed of multi-layer feature-plane and two basic operation modules: extended convolution and radially traversing probing. This engine can easily extract various image features such as moment, edge curvature, complexity, extent, blobs, and bars. We define“multiple cell-features”as a multi-dimensional feature which comprehensively represents properties of a subimage, which is here named a“cell”. The generalized Hough transform is introduecd as a universal method for object model matching using this multiple cell-features. This system can efficiently recognize objects unddr occlusion and noises. Model learning is performed by showing objects. In this paper, a system and hardware construction of vision system based on this engine is proposed. A prototype system demonstrates successful recognition of mechanical parts and equipment panels under uneven lighting-conditions and occlusion.
With a proper combination of compliant joint and position-controlled joint, a link system chatiges its posture with keeping contact between link system and environment. This is so-called Self-Posture Changing Motion (SPCM) and conveniently used to detect a contact point between robot and an unknown object. In conventional approach, a contact point has been computed as an intersecting point between two different link postures before and after SPCM. In general, this computation scheme gives us a contact point with unsatisfactory approximation, while it furnishes us with an exact contact point for a sharp-edged object. In this paper, we first introduce a concept of degree-of-approximation, which is defined as the distance between a computed point and the nearest point over the object surface, and then propose the Neighboring Degree-Of-Approximation Equating Method which improves the degree-of-approximation drastically and provides exact contact points not only for a sharp-edged object but also for a constant curved object. Mathematical framework is provided for the proposed computation scheme. Finally, the effectiveness of the proposed scheme is verified by simulations.
The aim of this paper is to clear characteristics of the obsever based robot motion controller, which observer estimates disturbance torque and velocity. Generally, because of perturbation for moment of inertia, gravity, friction, and low-stiffness by its kinematic structures, robot motion becomes worse. It is well-known that the disturbance observer based controller is robust for not only actually disturbance but unknown perturbation. But we have never full knowledge of that observer, today. This paper describes the phisical means of the disturbance torque and velocity estimated observer, at first. Next, it shows how the observer which is used as way of attenuation for additional disturbance, gives influence to not-nominal model. And, I will show the influence of observer and phase-shift for chaging moment of inertia. And for low stiffness manipulators, I will clear that the observer having not-optimal poles gives worse result for motion.
A new type of wall climbing robot, named Disk Rover, using permanent magnet disks are developed. The newly introduced permanent magnet disk is to rotate the magnet disk on the surface of wall with partly contacted posture. It allows to produce high magnetic attraction force compared with conventional permanent wheel which utilizes only a small portion of the magnet installed around the wheel. The optimum design of the magnetic wheel is done by using finit element method and it is shown that the magnetic attraction force vs. weight ratio can be designed about three times higher than conventional type magnet wheel. The developed Disk Rover is 25 kg in weight including controller and battery, about 685 mm in diameter, 239 mm in height and has a pair of permanent magnet disks. It is demonstrated by the experiments that the Disk Rover can move around on the surface of the wall quite smoothly by radio control and has payload of about its own weight. Several considerations are also done in order to surmount bead weld.
In proportion to the generalization of robots, tutorial on robotics plays an important role in university. This paper presents a basic system of instruction for robotics and manipulator. The system is intended to be applied to a students' lab. The authors have selected a compact three-joint manipulator for a tutorial program. The manipulator is designed to be an entry machine for robotics research, which enables easy handling and real-time calculation of kinematics. The tutorial program has three levels. Level 1 is a tutorial on basic servo system. In this course, the student is instructed in the step and frequency response characteristics. Study on nonlinearity is included in the course. Level 2 introduces the student to dynamic control. The emphasis of this course is on the effect of compensation for gravitational and inertial forces. Level 3 introduces the student to an application of the manipulator to a workspace. In this course, the student is instructed in PTP operation and trajectory planning. Performance of the system is evaluated by junior year students in University of Tsukuba. Through the experiments, the thought process of the students is investigated by protocol analysis.