Recent advancements in on-orbit space maintenance technologies have highlighted the limitations of manned spacecraft-based repair methods, particularly within the narrow space of satellites where dense component arrangements and severe spatial interference complicate maintenance operations. To address critical on-orbit tasks such as fault detection, repair, and component replacement, this study proposes an Adaptive-Step Heuristic Bidirectional RRT (ASH-BiRRT) algorithm optimized for narrow-space applications. Implemented on a six-degree-of-freedom (6-DOF) robotic manipulator platform, the algorithm achieves high-speed trajectory planning for maintenance manipulators. Validation experiments across obstacle scenarios of varying complexity demonstrate the method’s superiority. Extensive simulations confirm the algorithm’s high adaptability and exceptional efficiency in satellite cabin maintenance scenarios, establishing a robust framework for autonomous on-orbit maintenance in spatially constrained spacecraft interiors.