長期訪問割合目標を伴うLTL仕様に対する制御合成
Control Synthesis against LTL Specifications with Long-Run Visit Proportion Objectives
LTL仕様を満たしつつ、関心のある原子命題列の長期的な出現割合を指定範囲内に収める経路計画手法を提案し、四足ロボットで有効性を示した。
詳しい要約
1. どんなもの?
2. 先行研究と比べてどこがすごい?
3. 技術・手法の肝は?
4. どうやって有効だと検証した?
5. 議論はある?
6. 次に読むべき論文は?
※ AIが要旨から生成した要約です。正確性は原文をご確認ください。
著者: Zhiyuan Huang, Zhao Tong, Jiakai Li, Chenrui Xiang, Bingzhuo Zhong
分類: eess.SY
原文アブストラクト
This paper investigates the path-planning problem for systems required to satisfy a linear temporal logic (LTL) specification while achieving a desired long-run visit proportion. For a path represented in prefix-suffix structure, the long-run visit proportion quantifies the asymptotic occurrence proportion of an atomic proposition sequence of interest in the suffix trace. Such a quantitative requirement generally cannot be expressed by standard LTL specifications. Furthermore, we develop a planning approach that synthesizes an LTL-satisfying path whose long-run visit proportion remains within a prescribed tolerance of a desired value while satisfying an overall cost constraint. By adjusting the desired proportion, the synthesized path can allocate more or less long-run attention to the atomic proposition sequence of interest, thereby improving the flexibility and efficiency of the task execution. Finally, experiments on a quadruped robot demonstrate the practical significance of the proposed long-run visit proportion and the effectiveness of the proposed planning approach.