In this study, in order to examine the effect of shared experiences between humans and robots on the intimacy between the two, we examined changes in the robot's impressions by using a robot diary automatically generated from information during a joint experience. A robot diary was automatically generated by inputting the information obtained from the shared experience into a large-scale language model. To examine the effect of the proposed diary on the robot's impression, the robot and the researcher went for a walk and generated a diary including the interaction and dialogue history during the walk, and the experimental participants watched the walking video and evaluated the diary online. The results showed that the proposed diary made readers feel at ease significantly more in the robot's impression evaluation. In addition, when the impressions were examined for differences in diary content, the diaries with information during the shared experience were judged to be more favorable because they showed the robot's familiarity with the person from the robot. These results indicate that when a robot verbalizes and shares its own experiences with a person in a diary format, the intimacy between the two may be enhanced.

This paper proposes a speech synthesis model from multimodal information, i.e., text and lip movements, in order to generate more natural speeches including voiced and unvoiced sections. Its architecture consists of an image feature extractor using an auto-encoder and an encoder-decoder model that outputs a mel-spectrogram. Speech synthesis reflecting the lip movements to the text was successfully achieved. 3 types of combinations between text and lip movements were compared and evaluated.

In recent years, the development of Motion Platform has progressed, and movement methods in which the direction of body movement does not coincide with the video image have been proposed. However, little attention has been paid to the timing of movement switching. In this study, we examined the discrepancy in the timing of movement switching, and investigated a method of body movement that does not cause discomfort even when the timing is discrepant.

IMUs are used to estimate the self-position of automated vehicles, and in areas where satellites are blocked, the self-position is supplemented by information from the IMU's gyro. Therefore, it is essential to select a gyro with sufficient performance. However, the gyro evaluation method, Allan variance, is not applicable to dynamic environments because it is a geostationary indicator. Therefore, we propose a gyro evaluation method adapted to dynamic environments.

We propose a method that combines random sampling and machine learning for path generation of articulated robots in environments with obstacles. This method enables fast collection and training of data, as well as rapid and reproducible path inference, by training paths obtained through random sampling while dividing them into multi-layer perceptrons connected in multiple stages. Additionally, the effectiveness of this method is demonstrated through simulations.

The android robot applications for individuals with autism spectrum disorder (ASD) are investigated, especially for task-oriented dialogue such as job interview. Although android robot systems are potentially useful as conversational partners for individuals with ASD, their effectiveness has not been much considered. In this study, we develop an autonomous dialogue android robot system that can generate two types of utterances, context-aware or context-free, and adopt it for conversations with young adults with ASD. 14 young adults with ASD have chit-chat conversations with these two systems and assess them. The experimental results indicate that the preference for robot utterance type is different between participants, and the impression of contradiction could affect their preference.

To conduct naturally Teleoperated Bring-me tasks using a mobile manipulator, we assume an issue with difficulty in receiving an object happens when a user receives an object from the manipulator. In this paper, we first observe this issue in a Teleoperated Bring-me task using a Head Mount Display (HMD) and a mobile manipulator HSR. Then, we propose a novel immersive interface displaying a mirror-reversal viewpoint of the robot on an HMD to address the issue. Finally, we conducted a user experience experiment using the proposed interface compared to an interface displaying an original view of the HSR. We confirmed that the evaluation results increased the ease of the receiving task and operation of the HSR during the experiment.

A passenger of a personal mobility is required to drive in such a way as to avoid obstacles and deviations from the sidewalk. In this paper, therefore, we focus on driving assistance through shared control by both the mobility and the passenger. Online SLAM through sensor fusion allows the mobility to build a map composed of 3D point clouds with drivable and non-drivable areas. Furthermore, an artificial potential field, considering the repulsion from non-drivable areas and attraction by the passenger, is generated in the map. The mobility is controlled to move in the gradient direction of the potential field. As a result, the experiments show that the mobility based on the shared controller is able to move toward a destination while avoiding an obstacle ahead and deviations from a drivable area.