Unifying Foundation Models with Decision Making: A Path Towards Autonomous Self-Improving Agents beyond Rewards and Human Supervision