Visual Diagnostics for Deep Reinforcement Learning Policy Development
This work addresses the need for interpretability in safety-critical applications of reinforcement learning, though it is incremental as it adapts existing visualization methods to a new context.
The paper tackles the problem of interpreting vision-based deep reinforcement learning policies, which are often treated as black boxes, by extending CNN visualization algorithms to this domain, using a simulated drone environment to demonstrate how these tools can reveal policy qualities and flaws.
Modern vision-based reinforcement learning techniques often use convolutional neural networks (CNN) as universal function approximators to choose which action to take for a given visual input. Until recently, CNNs have been treated like black-box functions, but this mindset is especially dangerous when used for control in safety-critical settings. In this paper, we present our extensions of CNN visualization algorithms to the domain of vision-based reinforcement learning. We use a simulated drone environment as an example scenario. These visualization algorithms are an important tool for behavior introspection and provide insight into the qualities and flaws of trained policies when interacting with the physical world. A video may be seen at https://sites.google.com/view/drlvisual .