This paper is dealing with designing a state feedback controller without colliding so that the multivariable plant output is regulated not only to follow an output of the reference model but also to keep from an output of dynamical avoidance obstacle under following assumptions. (A.1) The plant control input is implemented on a basis of knowledge concerning parameters as well as a degree difference and concerning available measurements of states and inputs of plant, reference model and avoidance obstacle. (A.2) Each signal of reference model and avoidance obstacle inputs is produced without any knowledge on three systems and is not influenced by states and inputs of any other systems. (A.3) The obstacle output does not exist on a line between the plant and the reference outputs. Several features of the proposed controller are shown. (1) The controller is composed of safety following and avoiding controllers. (2) The gains of a safety following controller are designed from conditions not only to make the plant closed loop stable but also not to collide with the reference model output. (3) The gains of a safety avoidance controller are determined so that the plant closed loop is made unstable as well as the plant output does not collide with the avoidance obstacle output. (4) A switching function is introduced to adjust to switch the gains of safety following controllers to the ones of safety avoidance controller dependent on whether plant output meets across the prohibited region.
In this paper, we propose a method of cooperative transportation by two autonomous 4-legged robots. Since a legged robot's body vibrates seriously in walking, it is difficult to apply the same control method that are effective in cooperation of wheeled robots to legged robots. Then we made a flexible joint with elastic rubber to fix an object to the robot. It can sense force between the object and the robot itself. The two robots compose a leader-follower system and each robot controls its walking velocity and walking phase by its own information of force. We show experimental results of cooperative transportation based on the above method.
Analysis of positioning operations of multiple points on a deformable object and its planning problems will be discussed. In some operations that deal with deformable soft objects such as rubber parts, paper sheets, and textile fabrics, multiple points on the object should be guided to their desired locations. In addition, we cannot operate the guided points directly because of collisions among positioning devices and the other devices. Then we must realize the operation by controlling other points on the object. We call such operations indirect simultaneous positioning operations of multiple points on a deformable object and deal with these operations in this paper. We will propose positionability as the condition that positioned points can be guided to their given desired regions. Then, we will derive quantitative measures considering forces exerted on the object. Furthermore, planning method will be proposed based on the positionability condition and the proposed quantitative measures. Experimental results show the validity of the proposed planning method.
Recently, the needs of robot systems for human support such as a master-slave manipulator, a teleoperation robot and a power assist robot have been increasing. In such human-robot systems, a human operator has an initiative in executing a task, while a robot assists him or her. Therefore, the importance of a training system to improve operator's skill in controlling the robot should be recognized from a point of view of safety, since a control error of a human operator might cause a serious accident. In this paper, a new training system for human-robot systems is proposed, in which a neural network (NN) is used in order to identify the dynamic properties of the system and give an assist to the operator. The identification model used in the proposed system consists of the NN and a reference model which represents a control property of the skilled operator. This paper explains a working principle of the training method and shows the validity of the proposed method through experiments of robot control by novice operators.
In this paper, we develop a three DOF mechanism for humanoid robots, which we call the cybernetic shoulder. This mechanism imitates the motion of the human shoulder and does not have a fixed center of rotation, which enables unique human-like motion in contrast to the conventional design of anthropomorphic seven DOF manipulators that have base three joint axes intersecting at a fixed point. Taking advantage of the cybernetic shoulder's closed kinematic chain, we can easily introduce the programmable passive compliance adopting the actuation redundancy and elastic members. This is important for the integrated safety of humanoid robots that are inherently required to physically interact with the human.
In this paper, the bio-mimetic trajectory of robots for manipulating a holding nonholonomic car is generated with a Time Base Generator (TBG) . In order to reveal what kind of trajectories the robots should generate on the task, the experiments with subjects are performed, and it is shown that a human generates the trajectory with single or double-peaked velocity profile according to geometrical conditions of the car position. Then, by modeling these primitive profiles with the TBG, a new method that generates the bio-mimetic trajectory for the task of the robots is proposed. Finally, the trajectories are generated and compared with the human trajectories.
Most of existing robot learning methods assume that the environment where a robot works does not change, therefore, a robot has to learn from scratch if it encounters new environments. This paper proposes a method which adapts a robot to environmental changes by efficently transferring a learned policy in the previous environments into a new one and effectively modifying it to cope with the environmental changes. The resultant policy (a part of state transition map) does not seem optimal with respect to each individual environment, but may absorb the differences between multiple environments. We apply the method to a mobile robot navigation problem of which task is to reach the target avoiding obstacles based on uninterpreted sonar and visual information. Experimental results show the validity of the method.
Force-reflection is considerably essential in the complex teleoperations. This study aims to develop the practical force-reflecting teleoperators through the Internet, and supposes the time delay on the order of several hundred milliseconds. Since the transmission delays exist in network vary rapidly and possibly randomly, it is very difficult to ensure the stability. Accordingly, in this paper, the time-varying controllers which are adjusted to the round trip time delay observed on-line is designed based on the framework of the H∞ gain scheduling. This control strategy can achieve higher performance in the face of large variations in operating conditions. Furthermore, both communications protocol and compression of image data are investigated from the networked robotic point of view. The validity of proposed method is demonstrated by experiments of the bilateral teleoperation through the Internet.
It is expected that robots will work in homes and hospitals with humans in the future. We have developed a beach ball volley playing robot as a demonstration of human friendly robot technology. Our developed robot has the following three functions for a human friendly robot: 1) to interact with a robot using everyday words, such as “Let's play volleyball”, “Pick up a ball”, 2) to move under visual feedback by measuring a target position, and 3) to control arm motions which are expressed by using dynamic equations for performing tasks. This paper describes structures and procedures of the robot system.
This paper presents a method for path planning and motion execution of overhead travelling crane with three dimensional transport, considering the both of obstacle avoidance and suppression of sway for transfer objects. The present research is done based on a given geometrical map in environments for off-line application, as the basic step to establish autonomous mobile crane systems in future factory. A potential approach using a three-dimensional diffusion equation is applied to the present path planning. The moving direction and magnitude in each step are respectively calculated by the proposed vector-composite method and the allowable maximum values of acceleration and velocity constrained from the hardware capacity. Further, in order to avoid the collision with obstacles especially around the steep corner, a holding method is proposed by the introduction of the concept of a steering angle between the moving direction at the present position and the direction to be moved calculated from the potential field at the stopping-position of some-step ahead. The usefulness of this paper is demonstrated through computer simulations and laboratory experiments.
Robot, which has no actuators and walks down a slight slope, is called ‘passive walking robot’. It is well known that the passive walking robot is important and interesting target. In the past papers, , a stable steady periodic motion and a chaotic behavior of passive walking robot were analyzed by numerical simulations. To justify the numerical analysis, we have to show the occurrence through experiment. However, we have not ever seen experimental analysis concerned with those behaviors. Therefore, in this research, we constructed a passive walking robot named QUARTET II and tried to show the occurrence of the behaviors. In this paper, at first, we explain QUARTET II. Then, we show that, in passive walking, a few interesting behaviors mentioned above are observed via some numerical simulations. And finally, we show that those behaviors occur in walking experiments. Especially, it is noticed that we could observe a bifurcation in passive walking experiments. This is the main contribution of this paper. This experimental result is very important for biped robotics.
We are in the process of development of the“HELIOS-V”, which is equipped with 4 low-pressure tires outside and 2 high-pressure tires inside. This vehicle is designed to acquire the performance to ascend and descend stairs using low-pressure tires and move effectively on even ground by using high-pressure tires. To make the most of the characteristics of the vehicle, we have to understand the performance of the wheels in accordance with the terrain condition quantitatively. Therefore we constructed an experiment setup to evaluate performance of wheels on several terrain conditions without influence of the vehicle motion. By this experiment, we compared a low-pressure tire with a high-pressure tire and showed superiority of a low-pressure tire on rough terrain. We also evaluated 2 kinds of tires with slits and ribs to make grip force. Based on these considerations, we developed a Hard-Soft Wheel, or HS Wheel having both of the advantages of a low-pressure and high-pressure tire.
As one of the essential technologies to realize a demining robot, which has the function of searching task of the mines in autonomous phase and digging task of the mine in tele-operated phase, we propose a new type of master-slave arm system coined as “shape-feedback” master-slave arm. Feature mechanism of“shape-feedback”master and slave arm is the articu-lated finger joints located in the middle of master and slave fingers and their unilateral control to drive master finger joints to have same angle with slave finger joints, which bend with compliance. Addition of these finger joints under unilateral control on to the gripper with conventional bilateral servo control, it is expected to improve the performance of master-slave arm. We demonstrated the validity of the “shape-feedback” master and slave arm by the basic operati on experiments and experiments using the model of demining robot.