Multi-Objective Constraint Inference using Inverse reinforcement learning

TL;DR

This paper introduces MOCI, a framework for inferring shared constraints and individual preferences from heterogeneous expert demonstrations in reinforcement learning, improving accuracy and efficiency.

Contribution

MOCI is the first method to jointly infer shared constraints and individual preferences from diverse expert trajectories in reinforcement learning.

Findings

MOCI outperforms existing baselines in predictive accuracy.

MOCI maintains competitive computational efficiency.

MOCI effectively models diverse and conflicting behaviors.

Abstract

Constraint inference is widely considered essential to align reinforcement learning agents with safety boundaries and operational guidelines by observing expert demonstrations. However, existing approaches typically assume homogeneous demonstrations (i.e., generated by a single expert or multiple experts with identical objectives). They also have limited ability to capture individual preferences and often suffer from computational inefficiencies. In this paper, we introduce Multi-Objective Constraint Inference (MOCI), a novel framework designed to jointly extract shared constraints and individual preferences from heterogeneous expert trajectories, where multiple experts pursue different objectives. MOCI effectively models and learns from diverse, and potentially conflicting, behaviors. Empirical evaluations demonstrate that MOCI significantly outperforms existing baselines, achieving improved predictive performance, and maintaining competitive computational efficiency on a standard grid-world benchmark. These results establish MOCI as an accurate, flexible, and computationally practical approach for real-world constraint inference and preference learning tasks.