On-Site assembly is an important and complex task in many industrial fields, and it usually needs the cooperation of two skilled people. SuperLimb enhances the user's capabilities by increasing the number of limbs and provides the possibility for one person to perform on-site assembly tasks. However, explicit control from redundant body signals is difficult to control multiple degrees of freedom (DoFs) required for on-site assembly tasks due to limited control instructions. To address this issue, this paper proposes a Voluntary-Redundant hybrid control (VRHC) to achieve multiple DoFs control of SuperLimb by integrating the user's voluntary movement and redundant muscle signals. Nine intuitive redundant muscle patterns (IRMPs) are identified by a designed convolutional neural network and the muscle-velocity mapping method based on the redundant muscle force is proposed to control the velocity of the SuperLimb. The results show that the average accuracy of the IRMPs recognition is 94.03% and the average root-mean-square error with an ideal velocity-tracking curve of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\sin {x}$</tex-math></inline-formula> is 11.6% of the maximum velocity. Moreover, on-site assembly experiments are conducted to verify the coupling performance of redundant muscle control and voluntary movement. Consequently, the proposed method provides an intuitive velocity control with multiple DoFs for the SuperLimb to assist the user in complex coordination tasks.