# PEARL: Zero-shot Cross-task Preference Alignment and Robust Reward Learning for Robotic Manipulation

arXiv: https://arxiv.org/abs/2306.03615 · HTML: https://mlsanity.com/paper/2306.03615/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Placed in Reward modeling and RLHF, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 29 of its 45 references that are in the map.
- First posted on arXiv 6 Jun 2023.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Reward modeling and RLHF](https://mlsanity.com/alignment/reward-modeling-and-rlhf/)

## Builds on (most cited first, 29 of 29)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [Playing Atari with Deep Reinforcement Learning](https://mlsanity.com/paper/1312.5602/) (2013)
- [D4RL: Datasets for Deep Data-Driven Reinforcement Learning](https://mlsanity.com/paper/2004.07219/) (2020)
- [Offline Reinforcement Learning with Implicit Q-Learning](https://mlsanity.com/paper/2110.06169/) (2021)
- [Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learning](https://mlsanity.com/paper/1910.10897/) (2019)
- [Dota 2 with Large Scale Deep Reinforcement Learning](https://mlsanity.com/paper/1912.06680/) (2019)
- [What Matters in Learning from Offline Human Demonstrations for Robot Manipulation](https://mlsanity.com/paper/2108.03298/) (2021)
- [Tutorial on Variational Autoencoders](https://mlsanity.com/paper/1606.05908/) (2016)
- [Reward learning from human preferences and demonstrations in Atari](https://mlsanity.com/paper/1811.06521/) (2018)
- [PEBBLE: Feedback-Efficient Interactive Reinforcement Learning via Relabeling Experience and Unsupervised Pre-training](https://mlsanity.com/paper/2106.05091/) (2021)
- [DeepJDOT: Deep Joint Distribution Optimal Transport for Unsupervised Domain Adaptation](https://mlsanity.com/paper/1803.10081/) (2018)
- [Optimal Transport for structured data with application on graphs](https://mlsanity.com/paper/1805.09114/) (2018)
- [Data Uncertainty Learning in Face Recognition](https://mlsanity.com/paper/2003.11339/) (2020)
- [Gromov-Wasserstein Learning for Graph Matching and Node Embedding](https://mlsanity.com/paper/1901.06003/) (2019)
- [Primal Wasserstein Imitation Learning](https://mlsanity.com/paper/2006.04678/) (2020)
- [B-Pref: Benchmarking Preference-Based Reinforcement Learning](https://mlsanity.com/paper/2111.03026/) (2021)
- [Graph Optimal Transport for Cross-Domain Alignment](https://mlsanity.com/paper/2006.14744/) (2020)
- [SURF: Semi-supervised Reward Learning with Data Augmentation for Feedback-efficient Preference-based Reinforcement Learning](https://mlsanity.com/paper/2203.10050/) (2022)
- [Few-Shot Preference Learning for Human-in-the-Loop RL](https://mlsanity.com/paper/2212.03363/) (2022)
- [Preference Transformer: Modeling Human Preferences using Transformers for RL](https://mlsanity.com/paper/2303.00957/) (2023)
- [Teach a Robot to FISH: Versatile Imitation from One Minute of Demonstrations](https://mlsanity.com/paper/2303.01497/) (2023)
- [Inverse Preference Learning: Preference-based RL without a Reward Function](https://mlsanity.com/paper/2305.15363/) (2023)
- [Watch and Match: Supercharging Imitation with Regularized Optimal Transport](https://mlsanity.com/paper/2206.15469/) (2022)
- [Reward Uncertainty for Exploration in Preference-based Reinforcement Learning](https://mlsanity.com/paper/2205.12401/) (2022)
- [Cross-Domain Imitation Learning via Optimal Transport](https://mlsanity.com/paper/2110.03684/) (2021)
- [Optimal Transport for Offline Imitation Learning](https://mlsanity.com/paper/2303.13971/) (2023)
- [Reinforcement Learning from Diverse Human Preferences](https://mlsanity.com/paper/2301.11774/) (2023)

## Built on by (most cited first, 9 of 9)

- [GenPRM: Scaling Test-Time Compute of Process Reward Models via Generative Reasoning](https://mlsanity.com/paper/2504.00891/) (2025)
- [A Large Language Model-Driven Reward Design Framework via Dynamic Feedback for Reinforcement Learning](https://mlsanity.com/paper/2410.14660/) (2024)
- [RAT: Adversarial Attacks on Deep Reinforcement Agents for Targeted Behaviors](https://mlsanity.com/paper/2412.10713/) (2024)
- [VLP: Vision-Language Preference Learning for Embodied Manipulation](https://mlsanity.com/paper/2502.11918/) (2025)
- [Leveraging Sub-Optimal Data for Human-in-the-Loop Reinforcement Learning](https://mlsanity.com/paper/2405.00746/) (2024)
- [Progress Reward Modeling for Robotic Learning: A Comprehensive Survey](https://mlsanity.com/paper/2607.21655/) (2026)
- [RobotSmith: Generative Robotic Tool Design for Acquisition of Complex Manipulation Skills](https://mlsanity.com/paper/2506.14763/) (2025)
- [Similarity as Reward Alignment: Robust and Versatile Preference-based Reinforcement Learning](https://mlsanity.com/paper/2506.12529/) (2025)
- [Discovering Reinforcement Learning Interfaces with Large Language Models](https://mlsanity.com/paper/2605.03408/) (2026)

## Abstract (arXiv)

In preference-based Reinforcement Learning (RL), obtaining a large number of preference labels are both time-consuming and costly. Furthermore, the queried human preferences cannot be utilized for the new tasks. In this paper, we propose Zero-shot Cross-task Preference Alignment and Robust Reward Learning (PEARL), which learns policies from cross-task preference transfer without any human labels of the target task. Our contributions include two novel components that facilitate the transfer and learning process. The first is Cross-task Preference Alignment (CPA), which transfers the preferences between tasks via optimal transport. The key idea of CPA is to use Gromov-Wasserstein distance to align the trajectories between tasks, and the solved optimal transport matrix serves as the correspondence between trajectories. The target task preferences are computed as the weighted sum of source task preference labels with the correspondence as weights. Moreover, to ensure robust learning from these transferred labels, we introduce Robust Reward Learning (RRL), which considers both reward mean and uncertainty by modeling rewards as Gaussian distributions. Empirical results on robotic manipulation tasks from Meta-World and Robomimic demonstrate that our method is capable of transferring preference labels across tasks accurately and then learns well-behaved policies. Notably, our approach significantly exceeds existing methods when there are few human preferences. The code and videos of our method are available at: https://sites.google.com/view/pearl-preference.