The stable control of space robots after capturing noncooperative targets with unknown dynamic characteristics is challenging. Based on the adaptive reaction null space (ARNS) algorithm, the dynamic online planning path of a robotic arm was investigated to minimise the disturbance to the base. Using a variable data-window-size recursive least-square algorithm (VDW-RLS), the system could achieve an online estimation of the inertial parameters of the end link. In addition, a high-precision trajectory tracking control method was developed using the singular perturbation theory. According to the time scale, the system was decomposed into two subsystems. An elastic vibration compensator was designed for the rapid subsystem to suppress elastic vibration, and a robust adaptive controller was designed for the slow subsystem to track the adaptive reactionless trajectory. The validity of the adaptive reactionless path planning scheme via VDW-RLS and combined controller based on the singular perturbation technique were verified through simulation experiments.