<p>The model uncertainties are one of the factors affecting the reduction of estimation accuracy in computing lithium-ion battery charge levels. For this purpose, they use robust estimators to consider model uncertainties during the design process. Among the robust estimators, sliding mode observers are among the most widely used choices, which exhibit a chattering phenomenon during their performance, which consequently, decreases the estimation accuracy. To resolve this issue, the current paper designs an adaptive estimator from the sliding mode family, which largely eliminates the chattering phenomenon. In this solution, the gain of the estimator is adapted by a special dynamic based on the estimation error to remove the fluctuations caused by the sign function and as a result chattering phenomenon. In addition, to increase the convergence accuracy and also the estimation convergence speed, a Gary Wolfe optimization algorithm was employed for the optimization of the desired observer parameters. The effectiveness of the recommended strategy was checked using practical data, and after comparing the performance of this method to estimate the charge level with the performance of the conventional sliding method, it was established that this estimator could show good resistance against the increase of model uncertainties and it can provide an accurate assessment of the battery’s charge level.</p>