Mission-driven Exploration for Accelerated Deep Reinforcement Learning with Temporal Logic Task Specifications

TL;DR

This paper introduces a novel deep Q-learning algorithm that accelerates learning for agents with complex temporal logic tasks by using a mission-driven exploration strategy, improving efficiency in robot navigation scenarios.

Contribution

The paper presents a new deep Q-learning method that enhances sample efficiency through a mission-driven exploration approach based on automaton and neural network models.

Findings

Significantly faster learning in LTL-based control tasks

Improved sample efficiency over existing DRL algorithms

Effective in unseen robot navigation environments

Abstract

This paper addresses the problem of designing control policies for agents with unknown stochastic dynamics and control objectives specified using Linear Temporal Logic (LTL). Recent Deep Reinforcement Learning (DRL) algorithms have aimed to compute policies that maximize the satisfaction probability of LTL formulas, but they often suffer from slow learning performance. To address this, we introduce a novel Deep Q-learning algorithm that significantly improves learning speed. The enhanced sample efficiency stems from a mission-driven exploration strategy that prioritizes exploration towards directions likely to contribute to mission success. Identifying these directions relies on an automaton representation of the LTL task as well as a learned neural network that partially models the agent-environment interaction. We provide comparative experiments demonstrating the efficiency of our algorithm on robot navigation tasks in unseen environments.