The terminal centroid of the large component adjustment mechanism is an important parameter that affects its dynamic model. An inaccurate centroid position can cause the adjustment driving force to deviate from the ideal state, resulting in deformation of the adjustment mechanism and ultimately affecting the alignment accuracy of the assembly features. To meet the requirements for high-precision and compliant positioning of large-scale structures, this paper proposes a centroid position calibration method for the end of the posture adjustment mechanism that accounts for the friction torque of spherical joints. Firstly, a dynamic model of the posture adjustment mechanism is established, and the influence of centroid deviation on the internal force of posture adjustment is analyzed using numerical simulations. Secondly, based on the principle of force and torque transformation between coordinate systems, a centroid calculation model that incorporates the friction torque of ball joints is constructed. Meanwhile, the M-estimate sample consensus (MSAC) algorithm is adopted to optimize centroid parameters, so as to reduce the impact of random sensor errors and motion errors of positioners. Then, the Monte Carlo method is used to study the effects of random errors and centroid calibration strategies on calibration accuracy. The results show that sensor measurement error is the main random factor affecting the accuracy of centroid calibration. Increasing the maximum rotation angle of the terminal and the number of weightings can improve calibration accuracy, and the optimization effect of the MSAC algorithm is significant when the number of weightings is ≥ 6. Finally, centroid calibration of the mechanism terminal and posture adjustment alignment experiments of wing-body structures are conducted in a laboratory environment. The experimental results indicate that the proposed centroid calibration method can significantly improve the calibration accuracy of the terminal centroid. Using the centroid parameters obtained by this method for driving force planning, the average internal posture adjustment force is reduced by 56.0%, and the average coaxiality deviation of the fork ear holes is decreased by 47.6%, verifying that the proposed centroid calibration method can effectively reduce the internal forces of posture adjustment and improve the alignment accuracy of assembly features.