| DANG VAN TRONG | D, 中間発表 | ヒューマンロボティクス | 和田 隆広, | 松原 崇充, | 織田 泰彰, | 劉 海龍, | 本司 澄空 |
|
Title: A Cooperation Control Framework Based on Admittance Control and Time-varying Semi-passive Velocity Field Control for Human-Robot Co-carrying Tasks Abstract: Developing a cooperation control framework for human-robot co-carrying tasks in achieving safe interaction and completing the shared tasks, especially in scenarios where human intentions vary due to changes in the environment or task demands, remains an open challenge. Passivity-based control theory realizes safe operation by preserving the energetically passive relationship between the closed-loop system and its physical environment. However, solely adhering to passivity constraints may impose fundamental limitations on control performance and, in some cases, prevent the successful execution of controlled tasks. To address these issues, a cooperation control framework for human-robot co-carrying tasks was constructed by utilizing a reference generator and a time-varying semi-passive velocity field control in this study. Firstly, the human motion predictions are corrected in the event of prediction errors based on human-robot conflicts measured by the interaction forces through admittance control, thereby mitigating conflict levels. Subsequently, a time-varying semi-passive velocity field control approach is proposed, which utilizes the output of the motion generator to regulate robot behaviors during physical interaction with the human. In this manner, the proposed framework relaxes the inherently conservative nature in a controlled manner, which ensures the passivity of the closed-loop system when the energy level exceeds the designed one; and vice versa. Furthermore, the proposed control approach ensures that the system's kinetic energy is compensated within a finite time interval. The passivity, stability, convergence rate of energy, and power flow regulation are analyzed from theoretical viewpoints. Language of the presentation: English | |||||||
| LIU HUAKUN | D, 中間発表 | サイバネティクス・リアリティ工学 | 清川 清, | 和田 隆広, | 内山 英昭, | Perusquia Hernandez Monica, | 平尾 悠太朗 |
|
Title: Understanding Human Motion: From Sensing to Perceiving
Abstract: Understanding human motion is critical for enabling machines to perceive, interpret, and respond to human behavior in real-world scenarios. This work presents a unified research agenda spanning three interconnected directions: (1) motion estimation from sparse wearable sensors, (2) multi-person motion reconstruction, and (3) emotion perception from human body dynamics. We first investigate sensor configurations and propose model-based approaches to recover full-body motion from body-worn sensor measurements. Building upon this foundation, we extend the framework to multi-person settings, addressing spatial ambiguity and modeling inter-person interactions through range-based sensing. Finally, we explore how generative models can learn expressive motion priors to decode affective states embedded in posture and dynamics, advancing machine understanding of human emotion beyond facial or verbal cues. Language of the presentation: English | |||||||
| HOVHANNISYAN ANI | D, 中間発表 | ソフトウェア工学 | 松本 健一, | 飯田 元, | Raula Gaikovina Kula, | 嶋利 一真, | Fan Youmei |
Title:What Leads to External Documentation Churn? A Large-scale Study of Merge Requests on the Linux KernelAbstract:For large-scale codebases, maintaining up-to-date and clear documentation is essential for developers. Outdated documentation is a common issue for maintainers, especially when source code changes often lack accompanying documentation. As a large-scale codebase we target Linux kernel which evolves over the years and effective documentation management becomes necessary. We use documentation churn metric to analyse merge request (MR) subject, content, commit changes, modified file categories, and discussion comments in the Linux Kernel Mailing List, we aim to investigate the extent and types of MRs driving documentation churn. To achieve this, we analyzed around 250K MRs from 10 years patchwork of Linux kernel codebase. Results show merge requests mostly trigger Security related documentation churns at 23.5% rate of entire patchwork, and merge requests titled Documentation by maintainers are most likely to trigger "Device Drivers" and "User-Space Tools" subsystems, highlighting the need for targeted documentation management for other subsystems and proper linkage between MRs and churns. Our work is a large empirical study that shows evidence of different types of MRs that are more likely to be documented. We envision our work to assist developers with tools and techniques so that developers can better document their code changes. In the future, we plan to extend the dataset to other ecosystems, and understand to what extend content and discussions in MRs can translate to documentation churns using LLMs models.Language of the presentation:English | |||||||
| MUHAMMAD ALQAAF SUBANDOKO | D, 中間発表 | 計算システムズ生物学 | 金谷 重彦, | 松本 健一, | MD.Altaf-Ul-Amin, | 小野 直亮 | |
TitleIntegrative Computational Pipeline for Natural Product-Based Inhibitors of Viral Diseases AbstractBuilding on our published work against SARS-CoV-2, where we clustered 204 spike-glycoprotein sequences, matched 33,722 binding molecules to 52,107 secondary metabolites (SMs) from KNApSAcK, and identified fourteen promising natural inhibitors, we are now extending the same integrative pipeline to HIV-1. For the HIV-1 study, we first assembled an Ayurvedic NP library of ~3,200 unique SMs by mining KNApSAcK and hand-curated literature sources. We gathered FASTA sequences for all HIV-1 protein targets (protease, reverse transcriptase, integrase, envelope glycoprotein) from known protein targets of each SM in our Ayurvedic library. We then computed k-mer (k=3) frequency vectors for each FASTA, calculated pairwise similarity scores, and performed clustering using DPClusSBO to group proteins with similar sequence features. By mapping each SM to its cluster of protein targets, we can prioritize those SMs whose native targets cluster most closely with one or more HIV-1 proteins, thereby harnessing sequence-based "target homology" as a filter. Representative SMs from these clusters have now been selected for the upcoming docking stage against HIV-1 protease, RT, integrase, and Env. The resulting subset will proceed to in silico docking, Lipinski and bioavailability filtering for validation. LanguageEnglish | |||||||