This paper proposes a novel autonomous role allocation method for Multi-Agent Systems (MAS), inspired by the division of labor mechanisms observed in ant colonies. As MAS lack a centralized control architecture, global order must emerge through local interactions between autonomous agents and their environment. This inherent decentralized structure is expected to yield high robustness, flexibility, fault tolerance, and scalability. To address the challenge of autonomous coordination, we propose a role allocation framework incorporating three key components: the response threshold model, the response threshold variance model, and the response threshold specialization model. These models are derived directly from observed behaviors in ant colonies. Through computational simulations, we demonstrate the effectiveness of this approach in dynamic task allocation problems. Our results indicate that the proposed method enables agents to adaptively allocate roles in response to both internal and external system changes through local communication among agents.
Recently, balancing task efficiency and operator well-being has become increasingly important. For severely disabled operators collaborating with semi-autonomous, self-projected teleoperated robots, traditional support interventions prioritizing task efficiency (an efficiency-oriented design approach) may lead to decreased satisfaction and increased anxiety. In this study, we explore a feasible design methodology that integrates these two objectives—a flow-theory-informed design approach—by establishing a new design target prioritizing operator psychological well-being through the induction of flow states. Specifically, this study aims to compare flow-theory-informed design with efficiency-oriented design for clarifying the effects of each approach. To facilitate this comparison, we employed Kansei Engineering methods to construct a hierarchical model connecting conceptual targets (enhancing flow intensity) to concrete design elements, thereby elucidating the structural relationships among these approaches. The results revealed the existence of conflicting design elements between the flow-theory-informed and efficiency-oriented approaches, suggesting an underlying trade-off structure between them. Simultaneously, we identified synergetic design elements that benefit both approaches, as well as ambiguous elements regarding flow intensity. These insights provide valuable guidance for extracting elements that contribute to resolving the trade-off structure from the vast design space accumulated in teleoperation support research, offering a focused design direction for next-stage verification.
This paper presents an empirical study of position estimation for the small bioinspired robotic mouse TUCmouse. The objective was to integrate a low-cost inertial measurement unit (IMU) into the robot and evaluate whether approximate indoor position estimation can be achieved without external onboard localization sensors. Although established dead-reckoning and legged-robot state-estimation methods exist, applying them to a small, lightweight robot introduces additional challenges related to scale, sensor noise, body dynamics, and the lack of direct foot-contact sensing. Despite these challenges, this study evaluates three simplified approaches: orientation-aided step counting, foot-kinematic estimation, and an EKF-based method using IMU prediction with covariance-weighted kinematic and zero-velocity pseudomeasurements. Experiments on wooden and elastomeric polymer surfaces show that the EKF-based method achieves the lowest endpoint error among the tested methods. The results demonstrate both the potential and the limitations of low-cost inertial sensing for short-range approximate localization on small bioinspired robots.
Biohybrid approaches using robotic or virtual conspecifics have advanced the study and modulation of fish collective behavior. However, physical robots can introduce mechanical noise and model-dependent constraints, making it difficult to isolate visual effects. We developed a non-contact closed-loop projection system that generates manipulated self-shadows directly from the fish’s own movements without requiring a physical actuator or predefined kinematic model. Thirteen pairs of juvenile Plecoglossus altivelis were tested under three conditions: no-delay, delay (500 ms), and Flipped. Pair behavior was classified into three states based on inter-individual distance and wall proximity: co-wall, co-center, and split. Fish exposed to the manipulated self-shadows showed reduced wall-following while maintaining group cohesion. These findings suggest that temporally and spatially manipulated self-shadows can modulate fish collective behavior through visual information alone while preserving group cohesion.
Automated auscultation using wearable devices is essential for remote respiratory monitoring, but deep learning models often struggle to generalize due to the severe scarcity of annotated abnormal respiratory sounds. Since directly using synthetic data for supervised training risks learning artifacts instead of true pathological features, we propose a robust three-phase pipeline leveraging massive synthetic data without synthetic labels. First, a modified StyleGAN2 natively synthesizes rectangular Mel-spectrograms to preserve high-temporal-resolution acoustic characteristics, validated by kernel audio distance. Second, 100,000 synthetic spectrograms are used for unsupervised variational autoencoder pre-training, introducing a parallel asymmetric convolutional block to independently capture distinct time and frequency semantics. Finally, the encoder is repurposed as a feature extractor to train a lightweight classifier on limited clinical data. Empirical evaluations demonstrate that while a serial asymmetric kernel geometry yields the highest classification accuracy under frequency-dominant pathologies, our parallel architecture achieves highly competitive performance using only 43
