Safety Aware Reinforcement Learning (SARL)

TL;DR

This paper introduces SARL, a framework where a virtual safe agent learns a task-independent safety policy to modulate various primary agents, reducing side effects across different tasks without additional training.

Contribution

SARL proposes a task-independent safety modulation mechanism that can be applied to multiple policies, enhancing safety and generalizability in reinforcement learning environments.

Findings

SARL effectively minimizes side effects in complex dynamic environments.

The safety agent's modulation matches task-specific solutions in performance.

SARL demonstrates portability across different tasks without retraining.

Abstract

As reinforcement learning agents become increasingly integrated into complex, real-world environments, designing for safety becomes a critical consideration. We specifically focus on researching scenarios where agents can cause undesired side effects while executing a policy on a primary task. Since one can define multiple tasks for a given environment dynamics, there are two important challenges. First, we need to abstract the concept of safety that applies broadly to that environment independent of the specific task being executed. Second, we need a mechanism for the abstracted notion of safety to modulate the actions of agents executing different policies to minimize their side-effects. In this work, we propose Safety Aware Reinforcement Learning (SARL) - a framework where a virtual safe agent modulates the actions of a main reward-based agent to minimize side effects. The safe agent learns a task-independent notion of safety for a given environment. The main agent is then trained with a regularization loss given by the distance between the native action probabilities of the two agents. Since the safe agent effectively abstracts a task-independent notion of safety via its action probabilities, it can be ported to modulate multiple policies solving different tasks within the given environment without further training. We contrast this with solutions that rely on task-specific regularization metrics and test our framework on the SafeLife Suite, based on Conway's Game of Life, comprising a number of complex tasks in dynamic environments. We show that our solution is able to match the performance of solutions that rely on task-specific side-effect penalties on both the primary and safety objectives while additionally providing the benefit of generalizability and portability.