# PTRL：用于腿式机器人运动的先验迁移深度强化学习

> 原文链接：https://robotics.hellobike.com/papers/2026-08-01-2504-05629
> 频道：papers | 更新日期：2026-08-01 | 标签：arxiv, cs.RO
> 核心摘要：针对腿式机器人强化学习训练耗资源且难迁移，PTRL通过预训练、迁移、微调在多机器人行走任务上验证提升效率与性能。

# PTRL：用于腿式机器人运动的先验迁移深度强化学习

::: v-pre
> 原文链接：[PTRL: Prior Transfer Deep Reinforcement Learning for Legged Robots Locomotion](https://arxiv.org/abs/2504.05629)（arXiv 2504.05629 · 发布日期：2026-08-01）

**PTRL: Prior Transfer Deep Reinforcement Learning for Legged Robots Locomotion**

针对腿式机器人强化学习训练耗资源且难迁移，PTRL通过预训练、迁移、微调在多机器人行走任务上验证提升效率与性能。

## 核心要点

- 传统单机器人策略训练耗费大量计算资源，且模型专用复杂、难直接迁移应用。
- PTRL采用预训练、迁移、微调三阶段范式，并选择性冻结策略执行器关键层。
- 源机器人策略先基于PPO训练，再向目标机器人进行跨机器人知识迁移。
- 多机器人平台对比实验显示减少训练时间、提升行走性能，并揭示冻结比例与迁移效果关系。

**验证范围与局限**：验证范围为多机器人平台对比实验中的行走任务。

## 英文摘要 (Abstract)

Abstract Although reinforcement learning (RL) has shown great potential in legged robot motion control, traditional methods face two major bottlenecks: the strategy training of a single robot requires a large amount of computational resources, and the trained models are highly specific and complex for direct transfer and application. To address this key issue, this study proposes the Prior Transfer Reinforcement Learning (PTRL) framework, which improves training efficiency and model generalization ability through cross-robot knowledge transfer. The core contribution of this study is the construction of a three-stage learning paradigm of ‘pre-training, transfer, fine-tuning.’ Specifically, we first train the source robot strategy based on the Proximal Policy Optimization (PPO) algorithm and then achieve efficient knowledge transfer to the target robot by selectively freezing the key layers of the policy executor network. Through systematically designed multi-robot platform comparative experiments, it was verified that this method can reduce the training time, and the quantitative relationship between the proportion of frozen network layers and the transfer effect was revealed. The experimental results show that the proposed method significantly improves the performance of legged robots in walking tasks, demonstrating its strong applicability and advantages.

作者：Haodong Huang, Shilong Sun, Hailin Huang, Wenfu Xu
:::

## 原始参考来源 (Sources)

- [arXiv 2504.05629：PTRL: Prior Transfer Deep Reinforcement Learning for Legged Robots Locomotion](https://arxiv.org/abs/2504.05629)