Xinyu Cui

Fish can actively modulate the mechanical properties of their muscles to adapt to complex hydrodynamic environments, and extensive research has been dedicated to replicating such adaptive locomotion capabilities in bionic robotic fish. Unlike the widely investigated variable-stiffness tail designs in existing studies, this paper proposes a decoupled tunable rotational inertia mechanism for the robotic fish caudal peduncle, which enables independent regulation of tail rotational inertia without altering the global center of mass of the robot. A symmetric cable-driven transmission system equipped with a pair of oppositely moving mass spheres is adopted to realize real-time online inertia tuning while sustaining the robot’s dynamic balance. A high-fidelity fluid-structure interaction (FSI) dynamic model is established by integrating the pseudo rigid body method (PRBM) and the boundary data immersion method (BDIM). Simulation and experimental results demonstrate that rotational inertia modulation exerts a significant regulatory effect on the swimming performance of bionic robotic fish. Specifically, increased rotational inertia enhances thrust generation and forward swimming velocity within a specific tail-beat frequency range, whereas it degrades locomotion performance when the tail-beat frequency exceeds a critical threshold. Remarkably, the performance improvement enabled by enhanced rotational inertia can only be achieved via internal mass redistribution. In contrast, directly attaching additional mass to increase rotational inertia will almost invariably impair the swimming capability of the robot. This study reveals the unique regulatory effect of rotational inertia as an independent control input for bio-inspired underwater robots, providing a unique regulatory effect for the development of unified impedance-inertia adaptive control frameworks for future bionic underwater platforms.

Railway timetabling requires resolving complex scheduling conflicts arising from shared tracks, station capacity limits, and strict safety intervals. Existing optimization or learning-based approaches often struggle to scale or generalize across diverse operational scenarios. We present TrafSched, a novel hybrid decision framework that combines a curated strategy library, multi-dimensional conflict prioritization, Bayesian strategy adaptation and an optional Large Language Models (LLMs) integration module. TrafSched iteratively detects and resolves conflicts through adaptive strategy selection and backtracking, enabling robust exploration of feasible timetables without costly model retraining. Experiments on real-world-scale datasets involving 50–120 trains show that TrafSched consistently outperforms heuristic and reinforcement learning baselines, achieving up to 85.05% conflict-resolution success in the most challenging cases. These results demonstrate TrafSched’s effectiveness and scalability for modern railway scheduling operations.