LAUSR

Live-line maintenance robots operating at high altitudes facing significant vibration suppression challenges from wind load disturbances. To address this issue, this paper proposes a hybrid control method that integrates RBF neural networks and Zero-Moment Point (ZMP) theory. A three-dimensional ZMP stability reference surface model has been developed to overcome the adaptability limitations of traditional two-dimensional support criteria in flexible suspension scenarios. The method employs an adaptive online strategy using RBF neural networks. This strategy achieves coordinated optimization of global stability and local disturbance suppression by dynamically adjusting the ZMP stability domain. Numerical simulations under random wind loads (Beaufort scale 3-6) validate the method’s effectiveness. The proposed method reduces the steady-state ZMP tracking error to 5 mm. Compared with conventional RBF methods, it also achieves a 58.3% improvement in convergence speed and a 53.3% enhancement in postural stability. Finally, field deployment on live transmission lines confirms the technical feasibility of the method, significantly enhancing the safety of live-line operations. The research outcomes provide a novel theoretical framework and technical pathway for dynamic stability control in high-altitude robotic systems.