# Single-step Options for Adversary Driving

## TL;DR

This paper introduces a reinforcement learning approach using single-step options for adversary driving, which simplifies training and improves performance over primitive actions, human testers, and existing planning methods.

## Contribution

The paper proposes a novel single-step options framework that enhances training efficiency and outperforms existing methods in adversary driving scenarios.

## Key findings

- Faster and easier training compared to primitive-action agents
- Outperforms primitive-action RL, humans, and planning methods
- Demonstrates effectiveness of single-step options in adversary driving

## Abstract

In this paper, we use reinforcement learning for safety driving in adversary settings. In our work, the knowledge in state-of-art planning methods is reused by single-step options whose action suggestions are compared in parallel with primitive actions. We show two advantages by doing so. First, training this reinforcement learning agent is easier and faster than training the primitive-action agent. Second, our new agent outperforms the primitive-action reinforcement learning agent, human testers as well as the state-of-art planning methods that our agent queries as skill options.