The limited integrator anti-windup (LIAW) is a helpful method that can be applied to improve performance in systems with constrained inputs. However, the use of this technique has relied on no formal stability framework but on intuition. To the best of our knowledge, Lyapunov's theory has yet to be applied to study the advantages of the LIAW when used in an adaptive control scheme. This paper introduces an adaptive gravity compensation proportional-derivative (PD) plus LIAW algorithm to solve the joint position regulation of input-constrained robot manipulators. Moreover, this paper also reports the real-time experimental comparison among three control strategies: An adaptive nonlinear proportional-integral-derivative (PID) controller without any anti-windup extension, an adaptive PID scheme with a nonlinear anti-windup augmentation already reported in the literature, and the proposed adaptive PD plus LIAW algorithm. The experimental comparison departs from the idea of using PID-type controllers tuned with the same gains but without any special care for the input limitations. The advantages of the LIAW extension are then observed. Better performance results are obtained for the proposed adaptive PD plus LIAW solution.