Automatic person following method by independently tracking left and right feet parts is newly proposed. This method is applicable for small intelligent robot in near future, such as an automatic shopping cart. The proposed method consists of three parts. That is “Detecting and feature-learning of tracked person” part, “Searching lowest position of both feet” part, and “Robot control” part. The Condensation algorithm is utilized to robustly track both feet parts in conplex environment. Hypothesis (particles) are independently prepared for both left and right feet. Experimental results in indoor environment have shown the effectiveness of the proposed method.
This paper proposes a general optimal control algorithm for open-link mechanisms. Using the minimum principle, the optimal control problem can be transformed to a two-point boundary value problem of canonical equations of Hamilton. In order to solve the two-point boundary value problem efficiently, a recursive computational algorithm for canonical equations of Hamilton is developed based on the dynamics computational algorithm for link mechanisms. Then, a general optimal control algorithm for open-link mechanisms is formulated by connecting the present recursive algorithm and the robust optimal control algorithm called hierarchical gradient method that developed by authors previously. Some simulation results are shown to verify the effectiveness of the proposed algorithm.
This paper presents a robotic learning model for joint attention based on self-other motion equivalence. Joint attention is a type of imitation, by which a robot looks at the object that another person is looking at by producing an eye-head movement equivalent to the person's. It means that this ability can be acquired by detecting an equivalent relationship between the robot's movement and the person's. The model presented here enables a robot to detect the eye-head movement of a person as optical flow in the vision and the movement of its eyes and head as a motion vector in the somatic sense. Because both of the movements are represented with population codes, the robot can acquire the motion equivalence as simultaneous activations of homogeneous neurons that are responsible to a same motion direction in the two senses. Experimental results show that the model enables a robot to learn to establish joint attention based on the early detection of the self-other motion equivalence and that the equivalence is acquired in a well-structured visuomotor map. The results moreover provide analogies with the development of human infants, which indicates that the model might help to understand infant development.
This paper introduces a new light weight window cleaning robot with magnetic synchronous drive wheel and up/down wiper system. A pair of indoor/outdoor units attract each other by a magnetic force. Each unit has two slanted wheels to make a straight and curbed motion. Actuators are equipped only on the indoor unit, and the outdoor unit is a totally non electric system. A wheel on the outdoor unit is driven by a remote torque through a magnetic coupling where pair of multi-pole disk magnets are faced parallel on inside and outside of the glass. A wiper system also has a magnetic coupling to transmit an up/down motion from the indoor actuator to the outdoor linkages. Some experiments show the versatile locomotion and wiping performances.
This study discusses a feedback control system for reorientation of a planar space robot, whose angular momentum conservation leads to a nonholonomic constraint. One of the previous works for such systems defines a radially isometric orientation and establishes a smooth time-invariant feedback controller, but the proposed controller suffers from slow rate of convergence for a desired configuration placed near its zero-holonomy curve. This paper proposes a moving manifold based on a virtual desired configuration, which approaches to a real desired configuration in accordance with the distance to the moving manifold. The derived controller is effective for any desired configuration, and the convergence speed is improved. Some numerical simulations verify its effectiveness of the controller.
In this paper, the trajectory planning method of a semiconductor wafer transfer three-joint robot arm driven with stepping motors by use of the genetic algorithm is proposed, and the experimental results are demonstrated to confirm the usefulness of the present trajectory planning method. The whole trajectory consists of three trajectory portions; a straight line trajectory portion to take out the semiconductor wafer, a curved line trajectory portion under PTP action, and a straight line trajectory portion to set the semiconductor wafer. In the trajectory planning, the three trajectory portions are expressed by polynomials, and, by using the continuous conditions concerning joint angles, joint angular velocities and joint angular accelerations, the whole trajectory is described by a chromosome consisting of five genes. Then, the fitness function of the genetic algorithm for the quasi minimum time control under the constraint condition that the stepping motor torques should not exceed pull-out torques is defined, and the trajectory planning algorithm is constructed. Furthermore, the numerical calculations have been carried out, and it is confirmed that the trajectory planning can be successfully executed. Additionally, from the experimental results, it is ascertained that the trajectory tracking control of the trajectory of the semiconductor wafer transfer robot can be exactly implemented.
We aim at the development of rescue robots which are driven by ultra high pressure hydraulic small actuators. They are small-sized and have simple mechanisms, and can realize high-powered operations. This report shows the designs of two high powered small rescue robots-Jack Robot and Cutter Robot-and the results of the field tests carried out to verify in the possibilities. We carried out the field tests in the three types of field; collapsed house type, rubble type and traffic accident type. As the experimental results, the potential of high-powered small robots in disaster areas were shown.
This paper presents a feature extraction method for robotic motion learning that optimizes image resolution that is suitable to the task, thereby minimizing computation time. It utilizes mean-shift algorithms and principal component analysis for feature extraction, reinforcement learning for motion learning, and a trial and error algorithm for finding the appropriate resolution. The proposed feature extraction method was applied to a robot manipulation task. When applied to a manipulator pushing an object, the resolution adjustment method reduced the task time from approximately one minute to 21 seconds, which was caused by the appropriate selection of image resolution, while the performance of manipulation was maintained.
In this paper, we develop a prototype of a new snake-like robot using the screw drive mechanism, and design a control system for trajectory tracking. First, we explain the outline of the snake-like robot using the screw drive units and joint units, as well as the principle of the screw drive mechanism. Next, in order to achieve the trajectory tracking of the robot, we derive a kinematic model. Finally, the validity of the derived model and the effectiveness of the proposed control system are demonstrated by simulations and experiments.
A simple control method for redundant multi-joint arm was proposed recently by gaining a physical insight into human multi-joint reaching movements in redundancy of DOFs. Differently from the traditional approaches, the method need neither introduce any artificial performance index to resolve ill-posedness of inverse kinematics nor calculate the pseudo-inverse of the Jacobian matrix of task coordinates with respect to joint coordinates. This novel approach is based upon an idea of “Virtual spring-damper hypothesis, ” and the control signal is composed of linear superposition of three terms (1) joint-damping, and (2) virtual damper effects in parallel to (3) virtual spring effects in task space. This paper shows through experiments by using an industrial robot arm (PA-10) with redundant multi-joints that the control signal can generate smooth reaching movements without incurring any annoying selfmotion. It is shown further that virtual damper effects in task space enable endpoint trajectories of the robot arm to be rectilinear and the effects can be enhanced by compensating the viscosity inherent in the robot.
This paper describes a view-based localization method using support vector machines in outdoor environments. We have been developing a two-phase vision-based navigation method. In the training phase, the robot acquires image sequences along the desired route and automatically learns the route visually. In the subsequent autonomous navigation phase, the robot moves by localizing itself based on the comparison between input images and the learned route representation. Our previous localization method uses an object recognition method which is robust to changes of weather and the seasons; however it has many parameters and threshold values to be manually adjusted. This paper, therefore, applies a support vector machine (SVM) algorithm to this object recognition problem. SVM is also applied to discriminating locations based on the recognition results. In addition, to cope with image shifts caused by the variation of the robot's heading, we use a panoramic camera; we search the panoramic image for the region which matches the model image best. This two-stage SVM-based localization approach with a panoramic camera exhibits a considerable localization performacne for real outdoor image data without any manual adjustment of parameters and threshold values.