Robotic systems frequently operate under parametric uncertainties and constrained communication bandwidths, motivating data-driven control architectures that ensure optimal performance without explicit model identification. Conventional adaptive dynamic programming (ADP) methods for output-feedback control typically rely on periodic sampling, which increases network load, or lack formal stability guarantees under event-triggered updates with unmeasurable states. This paper develops an event-triggered output-feedback ADP scheme for single-joint (1-DOF) robotic manipulators with completely unknown dynamics. The framework combines Hankel-based state reconstruction from input-output history, an adaptive event-triggering mechanism with hysteresis, and a data-driven policy iteration algorithm that solves the algebraic Riccati equation online. Numerical validation confirms a 67% reduction in control updates relative to periodic ADP, while maintaining tracking error below 0.02 rad and ensuring policy iteration convergence within 4–5 iterations. Closed-loop uniform ultimate boundedness is proven with an explicit error bound , and Zeno execution is excluded via discrete-time Lipschitz analysis. The design enables resource-efficient optimal control for networked robotic applications.