Human-automation interactive approximate optimal shared control: A fixed-time learning approach

Effective collaboration between human operators and autonomous systems remains a critical challenge in modern cyber-physical systems. Although both human expertise and autonomous control demonstrate unique advantages, achieving seamless coordination under uncertainties and constraints presents significant theoretical and implementation difficulties. This paper proposes a novel fixed-time optimized shared control (FxT-OSC) methodology to enable efficient human-automation interaction through integrative learning. The main contributions include: 1) A game-theoretic adaptive confidence allocation mechanism that guarantees stable authority distribution between human and autonomous agents; 2) An innovative fixed-time (FxT) composite learning framework that accelerates optimal policy synthesis by combining human knowledge with machine learning; and 3) A comprehensive Lyapunov-based convergence analysis that provides explicit settling time bounds regardless of initial states. Simulation studies on unmanned aerial vehicle (UAV) attitude tracking validate the effectiveness of the proposed approach in achieving rapid learning convergence and precise control performance. The results demonstrate significant improvements in terms of learning efficiency and tracking accuracy.

Download paper here DOI

Recommended citation: Tan, Junkai and Ding, Zhenxiang and Guo, Zihang and Ren, Xuedong and Xue, Shuangsi and Cao, Hui (2025). Human-automation interactive approximate optimal shared control: A fixed-time learning approach. 2025 Joint International Conference on Automation-Intelligence-Safety (ICAIS) \& International Symposium on Autonomous Systems (ISAS).
Download Paper