This paper describes the design and development of a new robotic system in which the autonomous floor cleaning robot cooperates with the elevator to clean multiple floors in the high-rise office building without human intervention. The autonomous floor cleaning robot, which was reported in our previous paper, was modified and equipped with an optical transmitter to communicate with the elevator. The elevator accepts requests from the robot, opens the door and moves to the designated floor carrying the robot. The robot cleans specified area of the floor automatically and calls the elevator to move to the next floor when finished the cleaning task. We successfully introduced this system to a site of three office buildings. With only one human operator and seven robots, approximately 153, 600 [m2] in 80 floors can be cleaned in about 5 [h] and 30 [min] .
We would propose a novel mechanism of a surgical manipulator, which works in Open MRI guided surgery and assists surgeon (s) in precise positioning and handling of surgical devices. This mechanism can transmit 3 translational and 3 rotational motions from the outside to the inside of MR imaging area by using leverage and parallelogram mechanism. Such a remote actuation is significantly helpful because the strong and sensitive magnet of MR denies the accessibility of magnetic and electric devices and materials to the imaging area. In this paper, the basic ideas of mechanism were introduced and its forward and inverse kinematics was formulated firstly. Secondly, a non-MR-compatible prototype was made to reveal mechanical problems. Thirdly, the workspace of the prototype was numerically analyzed. Fourthly, the stiffness of the prototype was experimentally evaluated. These results concluded trade-off between stiffness and workspace. Fifthly, our new idea of workspace control for mechanical safety was preliminary discussed.
In this paper, we propose a novel tension calculation algorithm for tension-based force displays. The algorithm uses quadratic programming method to find optimal tension force and achieves stable force displays. We made some experiments to evaluate proposing method with a tension-based force display, SPIDAR. Results show that proposed method eliminates vibration and enlarges operating area.
This paper describes a novel fingertip equipped with a soft elastic cover and a solid nail. The elastic cover of the fingertip enables changing friction coefficient at the contact with an object. It also enables detecting change of contact type with an object: the fingertip contact with a flat surface and with a convex sharp edge. The solid nail enables detecting a very small step of sub-millimeter on a smooth surface.
Active Cord Mechanism is a mechanism in which the joints are coupled in many numbers in series, just like a snake. Typical studies of the propulsion of the Active Cord Mechanism focus on the 2D winding movement, like a snake. But Active Cord Mechanism may generate very efficient locomotion by introducing 3D motion. This paper proposes a new kind of propulsion principle for the underwater Active Cord Mechanism, in which each articulated body segment creates distortion motion, while keeping the whole body in a helical shape. The helical body will revolve around the center axis of the helices due to the forces generated by the distortion motion, and progress along the center axis of the helices. This paper also discusses about the design and control of a new mechanical prototype to realize a hermetic underwater 3D Active Cord Mechanism. Finally the effectiveness of the proposed principle for the underwater Active Cord Mechanism is evaluated by experiments using the developed robot.
This paper reports an experiment about interactive humanoid robots immersed in real human society with a com-munication task: language education. With recent robotics technology, a humanoid robot can provides an abundance of non-verbal information as well as vocal information, which enables us to smoothly communicate with it. We believe that these robots, partner robots, are advancing toward participation in human society. A humanoid robot named Robovie has been developed as such a partner to humans. It autonomously interacts with humans by speaking and making gestures with implemented child-like interactive behaviors. In addition, it employs a wireless system to identify individuals and adapt its behaviors to them. We performed an exploratory experiment to investigate the long-term relationships between humans and the partner robots that have a communication task. In the experiment, students in an elementary school interacted with two robots over 18 days. Each robot spoke in English with the Japanese students. We expected the students to form stable relationships with the robot by way of multi-modal communication and, as a result, improve their English language listening and speaking abilities. The experimental results show possibility of the partner robot in the language education; and they also provide considerable insights into developing partner robots that are well suited for immersion in human society.
In this paper, we propose a design method of task coordinated feedback control law for parallel link manipulators (PLMs) . In general, since it is the last purpose to control the coordinates of an End-effector by a manipulator's control, a task coofdinated feedback control suits the purpose rather than a joint coordinated one. However, control with a parallel link manipulator's work coordinates has the following two problems which should be solved. First, it is difficult for PLMs to have the information of task coordinate from joint infomation, though it is easy for serial link manipulators to do this. Second, it is necessary to calculate the Inverse Matrix of Jacobian for calculating Driving force of a joint. In this paper, we propose a novel solving method of forward kinematics problem for PLMs and construct a task coordinated feedback control which used the character of Jacobian and made calculation easy. Then we prove the stability of this control system and verify the validity by the simulation and experiment.
This paper presents balance control analysis of humanoid robot based on Zero Moment Point (ZMP) feedback control. The ZMP is often used as a standard evaluation of the stability of humanoid robot. Balance control is performed by controlling ZMP position so that it is always located in convex hull of the foot-support area. To simplify the design of controller, one-mass model which represents lower body part of the humanoid robot model has been commonly applied. However, the one-mass model causes the system becomes non-minimum phase, so that performance limitation is occurred. These disadvantages are attributed to the Waterbed effect in frequency domain and unavoidable undershoot in time domain. Therefore, this paper proposes the ZMP feedback control based on two-mass model, representing lower and upper body part of the humanoid robot. The proposed model results in minimum phase system. Finally, a design of controller based on the proposed model using Linear Quadratic Optimal Control by evaluating output of the system is described and confirmed using simulation.
Until present, it has been very hard to make land flat with teleoperated bulldozers through the video images from the cameras on the bulldozer. It has been assumed that the quality of images taken by the video camera was not sufficient for operators to get enough information of the work environment for teleoperation of the bulldozer. To clarify the possibility of teleoperation of the bulldozer through images from the video cameras on it, we compared the accuracies of the height and the angle of a bulldozer's blade under teleoperation and direct operation. Furthermore, we compared the accuracies of those under images from stereoscopic cameras which took the image of the horizontal index with the environmental image, and those under direct operation. As a result of the experiment, it was observed that it is possible to operate bulldozers more accurately through images from the stereoscopic camera system, which took the images of horizontal index with the environmental image, than through the images from the conventional stereoscopic camera system, even though the accuracy was worse than that under direct operation.
Reinforcement learning is very interesting for robot learning. However, there are some significant problems in applying conventional reinforcement learning algorithms to the robot with many degrees of freedom, because the size of exploration space increase exponentially with increase of degrees of freedom, and it makes it impossible to accomplish learning process. On the other hand, animals and humans can learn and accomplish various tasks using many redundant degrees of freedom of the body in spite of the exploration space is very huge. In this paper, we consider how to solve the state explosion problem in applying the reinforcement learning to the redundant robot and propose new framework of reinforcement learning, which is inspired by the body image of animals, by summarizing our previous works of reinforcement learning. To demonstrate the effectiveness of proposed method, simulations and experiments have been carried out and as a result effective behaviors have been obtained.