Cable-driven humanoid manipulators have gained widespread popularity in the human–robot interaction domain due to their lightweight design, flexibility, and safety. The working environments in human–robot interaction scenarios are typically unstructured, making obstacle avoidance path planning critical for cable-driven humanoid manipulators. However, the flexibility of cables and joint configurations result in low joint stiffness and low Cartesian stiffness for cable-driven humanoid manipulators, impacting operation accuracy and leading to potential failures in obstacle avoidance tasks. Based on the joint stiffness and Cartesian stiffness of the cable-driven manipulator, this work proposes a high-stiffness obstacle avoidance path planning algorithm incorporating a stiffness index for cable-driven humanoid manipulator. Finally, a prototype of a cable-driven humanoid manipulator is presented, and experimental results demonstrate that the proposed high-stiffness obstacle avoidance motion planning algorithm effectively improves the Cartesian stiffness performance of the cable-driven humanoid manipulator during obstacle avoidance motions.