Pretraining the Vision Transformer using self-supervised methods for vision based Deep Reinforcement Learning