Our new preprint Safe Continual Domain Adaptation after Sim2Real Transfer of Reinforcement Learning Policies in Robotics is out.