Many researches on lightweight robot link have been carried out. Lightweight robot arms will improve the energy consumption, maintainability, and safety. Such lightweight arms are likely to be flexible, so the control methods of these arms have been studied. On the other hand, using this flexible characteristic, it is pointed out that flexible arm can realize the tasks that cannot be done by conventional solid arms, when the arm is operated rapidly and large deformation on the arm is occurred. Using flexible arms, operations with high degrees of freedom can be achieved since they can maintain their shape or posture in stopping, and can change their shapes in moving. In this research, we focused on the manipulation method which uses the deformation of the flexible arm. We studied the system that is needed for controlling the deformation of flexible arms, and the method of extracting the deformation of the flexible arm from its entire shape. We conduct some experiments of oscillation suppression as an example of the controlling task. With the system constructed in this paper, it is confirmed that deformation of the flexible arm can be captured and that the controlling the flexible arm can be done.

We study a novel robot programming method that uses the view-based approach: “view-based teaching/playback.” This method directly uses images for robot programming and can adapt to changes of task conditions. In our previous work, it was successfully applied to simple force control tasks, wall-pushing, by visualizing force information with the photoelasticity. In this paper, we applied the view-based teaching/playback with photoelasticity to a more complex force-control task: wall-tracking. We used a robot finger equipped with a photoelastic fingertip and a grayscale camera with a polarized filter. In the experiment, successful wall-tracking was demonstrated in a target position that was different from those in teaching.

The goal of our research is to construct intelligence for autonomous robots which navigate through the real world using images from attached cameras. Localization is one of the key elements for autonomous navigation. In this paper, we propose a grid-based localization from a single image using deep learning. By using a grid map, uncertainty of localization can be defined in a natural way, which is useful for fusion with other sensors The network takes spherical panoramic images with position and orientation of the robot as training data. Position and orientation are expressed by multi-dimensional grid. The network behaves as a classifier, where each grid corresponds to indivisual class and its probability represents that of position and orientation. The experimental results support the effectiveness of the proposed method.

In this paper the authors introduce an intelligent system architecture for teleoperators, which is composed with networked software modules based on Robotics Technologies Middleware(RT-Middleware). The architecture consists of one hardware layer and two software layers. The first software layer can provide its task design function with modules for a user, and the user doesn't need to care about the actual structure and management of module connection in the system. This makes it possible to respond flexibly to the situation. Modules are hierarchically connected with each other virtually through the database node module (DNM). From the viewpoint of module load balancing, the methodology of autonomous module assignment is crucial for its efficient operation. We discuss about the steepest descent method for dynamic placement of the module.

We study a method for unknown object detection based on impacting and keypoint tracking. In this method, a robot changes object positions by impacting to detect each of the objects individually from camera images before and after impacting. This detection is possible because keypoints of each object always move consistently by impacting, while those of the background do not move. The proposed method is successfully applied to picking up of textured mahjong tiles by an industrial manipulator.

This paper proposed the flexible thermal management device: Reversible Thermal Panel (RTP). This device can control heat dissipation value passively depending on its thermal environment change and will be quite useful for the space probe which has wide range heat dissipation requirement. In order to propose the effectiveness of this device, this paper performs the design, fabrication, and experimental verification of the RTP system prototype. The prototype considers the realistic application and achieves larger heat dissipation compared with the previous researches. In order to achieve the huge heat dissipation, mechanical design of the RTP is completely renewed. The new design is confirmed through the numerical model simulation and experimental verification. After the design verification, the prototype system is tested under space environmental simulator. The test results show the variation of the heat dissipation value depending on the input power variation. In addition, the results show the possibility of the achievement of the real space probe heat dissipation requirement. The numerical analysis model is constructed using the test results. The way of the performance improvement is discussed with the help of the analysis model. Through the discussion of this paper, the proposed system shows the effectiveness and potential of the real space probe application.

This paper presents a novel methodology of introducing virtual mechanical elements for motion control of transferred objects in modeling of a five-axle, three-steering coupled-vehicle system. The coupled-vehicle system consists of two car-like mobile robots, two carriers and one steering unit, and it is capable of forming straight-bed and V-bed carriers depending on shapes of objects. In the kinematical model of the coupled-vehicle system, two virtual car-like mobile robots, a virtual carrier and two virtual revolute joints are newly introduced, although these virtual mechanical elements are redundant on converting the kinematical equations of the coupled-vehicle system into time differential equations in a chained form. The objective of introducing these virtual mechanical elements is to set the position of one of the virtual revolute joints and the orientation of the virtual carrier to be state variables in the chained form. Through the state variables, it is possible to specify the motion of an object placed on the V-bed carrier whose manipulation point is put on the virtual joint as the movement of the manipulation point and the rotation of the virtual carrier linked with the V-bed carrier around the point. Especially, the virtual revolute joint can be located inside a supporting triangle whose vertices are the three real revolute joints between the first car-like mobile robot, the first carrier, the second carrier and the second car-like mobile robot so that stable transportation is achieved when the manipulation point of the object is its center of mass. The validity of the control method based on the model in which the virtual mechanical elements are introduced is experimentally verified in transportation of an object placed on the V-bed carrier by following an 8th-order Bezier curve path while avoiding collision with an obstacle.