arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2604.04539cs.LGcs.RO
收藏

FlashSAC: 高维机器人控制中的快速稳定离线强化学习

FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control

KTH Royal Institute of Technology(皇家理工学院);German Research Center for AI (DFKI)(德国人工智能研究中心);Robotics Institute Germany (RIG)(德国机器人研究所)

Donghu Kim, Youngdo Lee, Minho Park, Kinam Kim, I Made Aswin Nahendra, Takuma Seno, Sehee Min, Daniel Palenicek, Florian Vogt, Danica Kragic, Jan Peters, Jaegul Choo, Hojoon Lee

AI总结:

FlashSAC基于Soft Actor-Critic提出快速稳定的离线强化学习算法,通过减少梯度更新并增大模型规模和数据吞吐量,在高维任务中超越PPO和强离线基线,显著提升训练效率和性能。

摘要:

强化学习(RL)是缺乏专家演示时机器人控制的核心方法。在线方法如近端策略优化(PPO)因其稳定性被广泛使用,但其依赖狭窄分布的在线数据限制了高维状态和动作空间中的准确策略评估。离线方法可通过学习更广的状态-动作分布来克服这一限制,但面临收敛慢和不稳定的问题,因为拟合价值函数需要大量梯度更新,导致通过靴値法积累批评误差。我们提出FlashSAC,一种基于Soft Actor-Critic的快速稳定离线RL算法。受监督学习中的缩放定律启发,FlashSAC大幅减少梯度更新,同时通过更大的模型和更高的数据吞吐量进行补偿。为在扩大规模时保持稳定性,FlashSAC显式限制权重、特征和梯度范数,抑制批评误差积累。在超过60个任务的10个模拟器中,FlashSAC在最终性能和训练效率上均优于PPO和强离线基线,尤其在高维任务如灵巧操作中收益最大。在模拟到现实的人形机器人 locomotion 中,FlashSAC将训练时间从小时减少到分钟,展示了离线RL在模拟到现实迁移中的潜力。

英文摘要:

Reinforcement learning (RL) is a core approach for robot control when expert demonstrations are unavailable. On-policy methods such as Proximal Policy Optimization (PPO) are widely used for their stability, but their reliance on narrowly distributed on-policy data limits accurate policy evaluation in high-dimensional state and action spaces. Off-policy methods can overcome this limitation by learning from a broader state-action distribution, yet suffer from slow convergence and instability, as fitting a value function over diverse data requires many gradient updates, causing critic errors to accumulate through bootstrapping. We present FlashSAC, a fast and stable off-policy RL algorithm built on Soft Actor-Critic. Motivated by scaling laws observed in supervised learning, FlashSAC sharply reduces gradient updates while compensating with larger models and higher data throughput. To maintain stability at increased scale, FlashSAC explicitly bounds weight, feature, and gradient norms, curbing critic error accumulation. Across over 60 tasks in 10 simulators, FlashSAC consistently outperforms PPO and strong off-policy baselines in both final performance and training efficiency, with the largest gains on high-dimensional tasks such as dexterous manipulation. In sim-to-real humanoid locomotion, FlashSAC reduces training time from hours to minutes, demonstrating the promise of off-policy RL for sim-to-real transfer.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑