Conservative DDPG offers a simple solution to the overestimation bias problem in RL without the need for ensembles.
