用少量数据训练出精准动力学模型，实现机器狗的高效抓取与运动控制。
Learning More With Less: Sample Efficient Model-Based RL for Loco-Manipulation
- 基于手工建模与贝叶斯神经网络，用少量数据快速学习精确动力学模型。
- 在波士顿动力Spot上实现动态末端轨迹跟踪，低数据量下仍保持高精度。
- 适合研究机器人控制、强化学习与少样本系统开发的学者与工程师。
将四足步行的灵活性与机械臂操作能力结合，使移动操作平台有望在真实场景中完成复杂任务。以波士顿动力Spot为代表的先进四足机器人已提供强大且稳健的平台。然而，移动操作控制的复杂性及商业平台的黑箱特性，给准确的动力学建模和鲁棒控制策略带来了挑战。为此，本文采用基于模型的强化学习方法。我们构建了一个带机械臂的四足平台的手工运动学模型，并利用近期发展的贝叶斯神经网络（BNN）技术，将其作为物理先验，从有限数据中高效学习准确的动力学模型。随后，基于所学模型通过强化学习推导出移动操作控制策略。我们在波士顿动力Spot硬件平台上验证了该方法的有效性，在低数据条件下实现了高精度的末端执行器动态轨迹跟踪。项目网站与视频：https://sites.google.com/view/learning-more-with-less。
原文摘要 · Abstract (English)
By combining the agility of legged locomotion with the capabilities of manipulation, loco-manipulation platforms have the potential to perform complex tasks in real-world applications. To this end, state-of-the-art quadrupeds with manipulators, such as the Boston Dynamics Spot, have emerged to provide a capable and robust platform. However, the complexity of loco-manipulation control, as well as the black-box nature of commercial platforms, pose challenges for deriving accurate dynamics models and robust control policies. To address these challenges, we turn to model-based reinforcement learning (RL). We develop a hand-crafted kinematic model of a quadruped-with-arm platform which - employing recent advances in Bayesian Neural Network (BNN)-based learning - we use as a physical prior to efficiently learn an accurate dynamics model from limited data. We then leverage our learned model to derive control policies for loco-manipulation via RL. We demonstrate the effectiveness of our approach on state-of-the-art hardware using the Boston Dynamics Spot, accurately performing dynamic end-effector trajectory tracking even in low data regimes. Project website and videos: https://sites.google.com/view/learning-more-with-less.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。