Safe Reinforcement Learning Using Robust Control Barrier Functions
This addresses safety challenges in RL for applications like robotics or autonomous systems, but it is incremental as it builds on existing safety-layer approaches.
The paper tackles the problem of unsafe exploration in reinforcement learning for safety-critical systems by proposing a differentiable robust-control-barrier-function safety layer in a model-based RL framework, demonstrating that it ensures safety and effectively guides exploration in experiments, including zero-shot transfer when rewards are learned modularly.
Reinforcement Learning (RL) has been shown to be effective in many scenarios. However, it typically requires the exploration of a sufficiently large number of state-action pairs, some of which may be unsafe. Consequently, its application to safety-critical systems remains a challenge. An increasingly common approach to address safety involves the addition of a safety layer that projects the RL actions onto a safe set of actions. In turn, a difficulty for such frameworks is how to effectively couple RL with the safety layer to improve the learning performance. In this paper, we frame safety as a differentiable robust-control-barrier-function layer in a model-based RL framework. Moreover, we also propose an approach to modularly learn the underlying reward-driven task, independent of safety constraints. We demonstrate that this approach both ensures safety and effectively guides exploration during training in a range of experiments, including zero-shot transfer when the reward is learned in a modular way.