Deep Reinforcement Learning is a promising tool for robotic control, yet practical application is often hindered by the difficulty of designing effective reward functions and training environments. For instance, Real-world tasks typically require optimizing multiple objectives simultaneously, necessitating precise tuning of their weights to learn a policy with the desired characteristics. Further, training tasks are often static and aim to learn challenging behaviors immediately from scratch. Instead, we propose using curriculum learning to progressively transition from easier tasks and objectives to more challenging and realistic ones. For that, we propose an integrated framework that trades off the learning difficulties of the tasks and objectives to optimally learn desired robotic behaviors. We aim validate our approach on various environments such as the Multi-Objective Gymnasium, ManiSkill3, and a mobile robot environment and compare to baselines trained without curricula. Main supervisor: Morteza Haghir Chehreghani, Department of Computer Science, Chalmers University of Technology and Gothenburg University