A Learnable Safety Measure
This work addresses safety constraints in reinforcement learning for physical systems, which is an incremental improvement over existing methods that require extensive prior knowledge.
The paper tackles the problem of ensuring safety in learning-based control of physical systems by introducing a learnable safety measure that captures the relationship between system dynamics and failure states, and demonstrates that using this measure during learning can greatly reduce failures.
Failures are challenging for learning to control physical systems since they risk damage, time-consuming resets, and often provide little gradient information. Adding safety constraints to exploration typically requires a lot of prior knowledge and domain expertise. We present a safety measure which implicitly captures how the system dynamics relate to a set of failure states. Not only can this measure be used as a safety function, but also to directly compute the set of safe state-action pairs. Further, we show a model-free approach to learn this measure by active sampling using Gaussian processes. While safety can only be guaranteed after learning the safety measure, we show that failures can already be greatly reduced by using the estimated measure during learning.