<p>Recent adaptive Differential Evolution (DE) variants utilize single-estimator techniques such as adaptive pursuit and Q-learning for online mutation strategy selection. However, the single estimator techniques are known to suffer from the disadvantage of maximization bias. This paper investigates the effect of maximization bias in some of the recent adaptive DE variants and proposes a new adaptive DE variant that integrates the Double Q-learning for online mutation strategy selection. The Double Q-learning is a double-estimator technique that is known for reducing maximization bias. The performance of the proposed algorithm is validated on the CEC 2021 benchmark optimization problems by comparing it to the recent state-of-the-art adaptive DE variants.</p>