For a robot to move several to dozens of meters, we have investigated a technique for measuring the true value and the robot position. To obtain the true value of the robot position, the establishment of an evaluation standard of the self-position estimation of an outdoor mobile robot is required. We have constructed the experimental system of position and posture measurement by using an augmented reality (AR) marker. Robot's self-position estimation have been evaluated through an experiment involving an outdoor mobile robot. In the experiments in an indoor environment, we confirmed that true-value measurement of several-millimeter resolution and the position and posture tracking of a moving object were possible. From the experiments in an outdoor environment, we confirmed the automatic detection of the AR marker with a resolution of 3[mm]. We also confirmed that such marker detection and tracking of the robot position are possible for all frames in the moving image.

In this paper, we analyze the optimal placement of pulleys for multi-pulley wire transmission mechanisms designed to reduce sliding friction in wire-driven robotic joints. By optimizing the pulley arrangement, we maximize the range of joint motion. Additionally, we propose joint drive mechanisms for the quad-pulley wire transmission mechanism, demonstrating that wire tensions generate greater torque compared to conventional mechanisms. The effectiveness of the mechanisms is validated through experimental results, demonstrating reduced wire friction and analyzed range of joint motion. This study contributes to the development of robots capable of effective operation in uncertain environments by clarifying their mechanical design and functionality.

Self-reconfigurable modular robots (SRMRs) are capable of rapidly switching configurations to meet a diversity of applications and environments. To conduct automatic docking between different modules is essential for the self-assemble and self-transformation of modular robots. In this work, we have developed a method to conduct pose-estimation considering various information as well as perform accurate and robust docking between modules. Finally, we developed a modular robot simulator based on Isaac Sim and evaluated the effectiveness of the proposed method.

To improve the lifelikeness of the robot, we developed a bird-like robot that can reproduce egg laying, the birth of life. In this study, we conducted experiments with 12 pairs of parents and children to examine how the robot's egg laying affects the perception of parents and children about lifelikeness and mind of the robot, and how the parents and children feel towards the robot. The results showed that robot egg-laying leads to a stronger sense of mind perception of the robot. It was also shown that children felt a stronger sense of lifelikeness and trust in the robot than their parents.

This study explores the application of Dynamic Mode Decomposition (DMD), a data-driven modeling technique, for drone modeling. Conventional model-based control methods for dynamic systems, such as drones, require precise models, which are difficult to create due to challenges in obtaining exact component parameters and accounting for real-world disturbances. To overcome these issues, DMD with Control (DMDc), an extension of DMD to control systems, is applied to drone modeling. We collect time-series data from drones following circular trajectories and use this to construct a DMDc model. The model's accuracy was then evaluated on both circular and figure-8 trajectories.

In this paper, we propose the principle of the robot hand equipped that is equipped with belt mechanisms on its fingertips and uses two types of sensors in combination: a thin film pressure sensor mounted inside the belt mechanism, and a force sensor incorporated in the finger opening and closing drive system, to achieve wide-range force sensing, thereby enabling the hand to not only grasp objects ranging from soft to heavy, but also to pull-in and rotate those objects within the hand. Then, we discuss the configurations of the real machine, and the basic performance of force sensing from basic experiments.

In this study, we proposed an underactuated robot gripper that uses a single air cylinder to approach an object to be grasped and then grasp it. When the air cylinder is extended, the gripper moves toward the object to be grasped and switches to a grasping operation upon contact with the object. On the other hand, when the air cylinder is retracted, the object is released first, and then the gripper is pulled back to its initial position. In this paper, we built a prototype of the proposed gripper and conducted verification experiments. As a result of the verification, we confirmed that it is possible for the proposed gripper to approach an object to be grasped, to switch the gripper to a passive grasping operation, and to pick and place the object.

This paper presents a novel asynchronous vibrational orbit generation method for a vibrational actuator utilizing a flexible beam. High-frequency 2-dimentional Lissajous curve-like vibrational orbit can be generated by low-frequency 1-dimentional impulse train input. First, we propose an analytical model for the actuator and an impulse train input for calculating the orbit. Next, we demonstrate that by the impulse input with a period based on the common multiple of two terms of resonant frequency ratio, Lissajous curve-like orbit can be generated. Even if both resonant frequencies are higher than the input frequency, Lissajous curve-like orbit at the resonant frequencies can be generated. Finally, the proposed method is validated by experiment.