Learning-based control: PPO, SAC, behavior cloning, a diffusion policy and residual RL, trained offline on these plants and run here beside the model-based laws.