发表机构
Tampere University; Faculty of Engineering and Natural Sciences（坦佩雷大学; 工程与自然科学学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文针对受未知扰动的输入受限欠驱动线性系统，提出鲁棒安全滤波框架，结合H-∞控制、扰动观测器与控制障碍函数，通过两轮自平衡机器人仿真验证约束竞争问题。
AI 中文摘要
针对受未知扰动影响的输入受限欠驱动线性系统，本文提出一种鲁棒安全滤波框架。从零和微分博弈中推导得到基线H-∞输入，同时由扰动观测器提供扰动估计值及暂态误差界。利用扰动估计值调整基线输入，而估计值及其误差界则用于定义鲁棒高阶控制障碍函数约束；只要容许输入集保持非空，前向不变性即可成立。对于标量输入系统，逐点可行性由精确输入区间确定，区间宽度定义了可行性裕度。有限时域H-∞性能平衡则考虑施加输入与基线H-∞策略的累积偏差。对线性化两轮自平衡机器人的仿真表明，位置与车身俯仰约束会争夺同一受限的车轮转矩输入。
英文摘要
We develop a robust safety filter for input-constrained underactuated linear systems subject to unknown bounded-rate disturbances and structured model uncertainty entering through known distribution channels. A disturbance observer provides an online disturbance estimate and a dynamic estimation-error radius that define the certified uncertainty envelope used in the robust high-order control barrier function constraints. For general polytopic actuator sets, the resulting robust safety-admissible input set is polyhedral, while the shared scalar-input specialization yields an exact interval, a necessary-and-sufficient pointwise feasibility condition, a signed feasibility reserve, and a closed-form safety projection. A domain-wise certificate relates robust HOCBF control demand to available actuator authority, while a finite-horizon energy identity quantifies deviation from the unconstrained $H_\infty$ reference without claiming preservation of its original attenuation level. Across all $16$ structured-uncertainty corners in the linear-model actuator-stress test, the filter remains feasible and safe under $τ_{\max}=1.18$ N$\cdot$m, with $\max\abs{p}=0.0740$ m, $\max\absθ=0.205$ rad, $\max\absτ=1.173$ N$\cdot$m, and $\minμ_S=7.27\times10^{-3}$ N$\cdot$m.
