Semi-Supervised Cross-Domain Imitation Learning

TL;DR

This paper introduces a semi-supervised approach to cross-domain imitation learning that leverages limited expert demonstrations and unlabeled trajectories to improve policy transfer across domains.

Contribution

It proposes the first algorithm for semi-supervised cross-domain imitation learning with theoretical support and a novel loss function for domain discrepancy handling.

Findings

Achieves stable, data-efficient policy learning with minimal supervision.

Outperforms baseline methods on MuJoCo and Robosuite tasks.

Demonstrates effectiveness of the proposed cross-domain loss and adaptive weighting.

Abstract

Cross-domain imitation learning (CDIL) accelerates policy learning by transferring expert knowledge across domains, which is valuable in applications where the collection of expert data is costly. Existing methods are either supervised, relying on proxy tasks and explicit alignment, or unsupervised, aligning distributions without paired data, but often unstable. We introduce the Semi-Supervised CDIL (SS-CDIL) setting and propose the first algorithm for SS-CDIL with theoretical justification. Our method uses only offline data, including a small number of target expert demonstrations and some unlabeled imperfect trajectories. To handle domain discrepancy, we propose a novel cross-domain loss function for learning inter-domain state-action mappings and design an adaptive weight function to balance the source and target knowledge. Experiments on MuJoCo and Robosuite show consistent gains over the baselines, demonstrating that our approach achieves stable and data-efficient policy learning with minimal supervision. Our code is available at~ https://github.com/NYCU-RL-Bandits-Lab/CDIL.