Shield Decomposition for Safe Reinforcement Learning in General Partially Observable Multi-Agent Environments

Daniel Melcer, Christopher Amato, Stavros Tripakis

Keywords: shielding, multi-agent, safety, partial observability

TL;DR: Extend multi-agent shielding to work in general partially observable environments, making use of a SAT encoding of the problem

Abstract:

As Reinforcement Learning is increasingly used in safety-critical systems, it is important to restrict RL agents to only take safe actions.
Shielding is a promising approach to this task; however, in multi-agent domains, shielding has previously been restricted to environments where all agents observe the same information.
Most real-world tasks do not satisfy this strong assumption.
We discuss the theoretical foundations of multi-agent shielding in environments with general partial observability and develop a novel shielding method which is effective in such domains.
Through a series of experiments, we show that agents that use our shielding method are able to safely and successfully solve a variety of RL tasks, including tasks in which prior methods cannot be applied.

Submission Number: 254