arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2002.02991cs.ROcs.LGstat.ML
收藏

为类人机器人学习全身运动技能

Learning Whole-body Motor Skills for Humanoids

The University of Edinburgh(爱丁堡大学)

Chuanyu Yang, Kai Yuan, Wolfgang Merkt, Taku Komura, Sethu Vijayakumar, Zhibin Li

摘要:

本文提出了一种用于深度强化学习的分层框架,该框架可以获得各种推动恢复和平衡行为的运动技能,即,踝关节,髋关节,脚倾斜,和步进策略。该策略在物理模拟器中进行训练,该物理模拟器具有真实的机器人模型设置和低级别的阻抗控制,易于将学到的技能转移到真实的机器人上。与传统方法相比的优点是将高级规划器和反馈控制全部集成在一个单一的连贯策略网络中,这对于学习针对任意位置处的未知扰动的通用平衡和恢复运动是通用的(例如,腿、躯干)。此外,所提出的框架允许的政策,快速学习的许多国家的最先进的学习算法。通过比较我们的学习结果的预编程,特殊用途的控制器在文献中的研究,自学技能是可比的干扰抑制方面,但具有额外的优势,产生广泛的自适应,通用和鲁棒的行为。

英文摘要:

This paper presents a hierarchical framework for Deep Reinforcement Learning that acquires motor skills for a variety of push recovery and balancing behaviors, i.e., ankle, hip, foot tilting, and stepping strategies. The policy is trained in a physics simulator with realistic setting of robot model and low-level impedance control that are easy to transfer the learned skills to real robots. The advantage over traditional methods is the integration of high-level planner and feedback control all in one single coherent policy network, which is generic for learning versatile balancing and recovery motions against unknown perturbations at arbitrary locations (e.g., legs, torso). Furthermore, the proposed framework allows the policy to be learned quickly by many state-of-the-art learning algorithms. By comparing our learned results to studies of preprogrammed, special-purpose controllers in the literature, self-learned skills are comparable in terms of disturbance rejection but with additional advantages of producing a wide range of adaptive, versatile and robust behaviors.

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。

↑