We report the results of field study that qualitatively reveal the usefulness of collaborative researcher-robot presentations in a science museum from the viewpoints of (1) understanding the general public's thoughts on robots for children, (2) conveying information on robotic literacy to the general public, and (3) creating researcher-people interaction while walking together with a robot after a presentation. In robotics research for children, researchers need to deepen their understanding of the general public's expectations and concerns about robots, and the general public needs to deepen their understanding of the capabilities and limitations of robots. We believe that a collaborative researcher-robot presentation in a science museum allows for these, and also has the potential to generate discussions between a researcher and the general public in other scenes than presentations, through walking together with a robot. In this study, the first author and a robot conducted 16 series of science communication activities on robotics research in a science museum, in which more than 433 people participated in total. We evaluated these practices based on the first author's participant observation. We confirmed the usefulness of collaborative researcher-robot presentations in a science museum from the three perspectives, and also obtained unexpected findings that may have been obtained thanks to the first author's participant observation in the field. Our findings should be useful in conducting discussion activities with the general public regarding the research and development of robots that interact with children, and in considering how to show mobile robots in public spaces in a friendly manner.
With the development of deep learning, the recognition performance of automatic speech recognition has been greatly improved. On the other hand, there is still a problem of degradation of recognition accuracy due to an increase in the number of false positives of words and speech parts when environmental noise is severe. To solve this problem, many methods have been proposed to suppress the noise and to emphasize only the target speech, i.e., speech enhancement. In most cases, speech enhancement requires some assumptions to be made about the sound source. In addition, conventional speech enhancement methods do not fully utilize the key features in the input signal because they use a single model or network to enhance the speech. In this paper, we report a speech enhancement method based on beamforming using an ensemble time-frequency mask. The ensemble time-frequency mask is generated by estimating and integrating multiple time-frequency masks from multiple speech enhancement methods. The use of time-frequency masks estimated from multiple methods is expected to improve the robustness of the process. We evaluated the proposed method on the CHiME-3 dataset using PESQ and STOI, which are correlated with human auditory perception. In both evaluation metrics, the proposed method outperforms the one without ensemble, indicating the effectiveness of the proposed method. In addition, we conducted a validational experiment on the ensemble method of the proposed method.
Synergy Hand is a synergy-based multi-fingered hand that can express various postures with limited control inputs. Conventional synergies compressing the dimension of hand's motion by using the external characteristics of various grasping postures ruin the hand's dexterity. This is because functional information that each finger should perform is lost. This study proposes a synergy hand that can select the joints for synergy-based control by utilizing a fluid network. The proposed hand can switch its synergy according to the required task by switching the fluid path and can perform various tasks with limited control inputs. Our experimental results show that the proposed hand can grasp and manipulate various objects with 2 or 3 control inputs, respectively.
This paper presents the design and experiment of the instantaneous force generation mechanism based on the motion of the mantis shrimp, considering the occurrence of cavitation. This study aims to develop a robot that strengthen the striking force by generating cavitation. In conventional studies, it has been confirmed that mantis shrimp generated two consecutive impacts on a hard object by hitting and cavitation in water. However, there is no robot that can impact twice with a mantis shrimp's striking mechanism. In this paper, striking mechanism of mantis shrimp was modeled and the mechanism was designed and developed based on the model. From the experiment, it was confirmed that the proposed mechanism caused cavitation only at the point of collision with the object. This result indicates the possibility of strengthening the striking force.
Small rovers are suitable for carpooling or precursor missions and have been expected to explore complex areas where large rovers are hard to traverse. However, the moving performance of small rovers is lower than that of large rovers in general. This study designs a novel jumping mechanism that mimics a grasshopper leg for small rovers. We produced nine different leg expansion trajectories by changing the joint positions of the mechanism and investigated how these trajectories affected jumping performance. The highest jumping was obtained with a straight expansion trajectory and a kicking behind the foot, and kicking in front of the foot with a upward convex expansion trajectory generated the longest jump distance.
This paper presents a multi-legged rock-climbing robot for surface exploration of rugged, steep terrains, such as a cliff wall. For a robot hand toward robotic rock-climbing, passive grippers with sharp spines at the tips of their fingers have been developed. The grippers need to be locally-adaptable to micro-scale roughness of the uncertain terrain and to exert sufficient grasping forces for climbing. In this paper, we propose two grasping force compensation functions for the gripper, tracing grip and tolerance to drawing back the gripper, using a force sensor attached to the wrist part. These functions can achieve the adaptability to the grasping the uncertain rugged terrain surfaces. The effectiveness of the proposed compensation functions was verified through vertical climbing experiments of a teleoperated four-legged rock-climbing robot system, which was newly designed and developed, in a rugged wall. The results show that the developed robot system is capable of the vertical climbing in the rugged wall by re-grasping in response to the compensation based on the sensing data of the grasping state.
This research explores how robots can be used to protect parties to a conflict and civilians from suicide bombings in conflict areas under the international humanitarian law. This paper examines the issue from the perspectives of robotics and international law using a real robot. In this paper, the articles of international humanitarian law are interpreted, and a framework required for robotic systems is proposed. From the perspective of the principle of distinction, the principle of proportionality, and the principle of precaution, ideas are presented to calculate the legal indicators named “certainty,” “effort,” “nexus possibility,” “necessity,” and “collateral damage.”