Random forest is an extension of decision trees. It is a forest as it is a combination of multiple trees, and is random as we randomly sample different observations for each of the decision trees.
A random forest works by averaging the prediction of each of the decision trees (which work on a sample of the original dataset).
Typically, a random forest works better than a single decision tree, as the influence of outliers is reduced in it (because in some samples, outliers might not have occurred), whereas, in a decision tree, an outlier would have definitely occurred (if the original dataset contained an outlier).