Safe Planning and Policy Optimization via World Model Learning
This addresses safety and reliability issues in RL for real-world scenarios, representing an incremental improvement with novel mechanisms for adaptive planning and safety thresholds.
The paper tackles the problem of ensuring safety in reinforcement learning for real-world applications by proposing a model-based RL framework that jointly optimizes task performance and safety, achieving robust performance on diverse safety-critical continuous control tasks and outperforming existing methods.
Reinforcement Learning (RL) applications in real-world scenarios must prioritize safety and reliability, which impose strict constraints on agent behavior. Model-based RL leverages predictive world models for action planning and policy optimization, but inherent model inaccuracies can lead to catastrophic failures in safety-critical settings. We propose a novel model-based RL framework that jointly optimizes task performance and safety. To address world model errors, our method incorporates an adaptive mechanism that dynamically switches between model-based planning and direct policy execution. We resolve the objective mismatch problem of traditional model-based approaches using an implicit world model. Furthermore, our framework employs dynamic safety thresholds that adapt to the agent's evolving capabilities, consistently selecting actions that surpass safe policy suggestions in both performance and safety. Experiments demonstrate significant improvements over non-adaptive methods, showing that our approach optimizes safety and performance simultaneously rather than merely meeting minimum safety requirements. The proposed framework achieves robust performance on diverse safety-critical continuous control tasks, outperforming existing methods.