从跨形态演示中学习机器人操作
Learning Robot Manipulation from Cross-Morphology Demonstration
AI总结:
本文提出模仿学习中的形态适应框架MAIL,利用形态显著不同的智能体的次优演示训练机器人视觉控制策略,在刚体与布料操作任务中相比基线提升最高24%,并成功部署到真实Franka Panda机器人。
摘要:
一些从演示中学习(LfD)的方法可以处理教师与学生动作空间中的小幅度不匹配。本文针对教师形态与学生形态存在显著差异的情况进行研究。我们提出的框架——模仿学习中的形态适应(MAIL)——弥合了这一差距,使我们能够利用来自形态显著不同的其他智能体的演示来训练一个智能体。只要这些演示能为期望的解决方案提供一定的指导,MAIL 就能从次优演示中学习。我们在刚体和可变形物体的操作任务上展示了 MAIL,包括与刚性障碍物交互的三维布料操作。我们利用来自具有两个末端执行器的模拟智能体的演示,为具有一个末端执行器的机器人训练视觉控制策略。在归一化性能指标上,MAIL 相比 LfD 和非 LfD 基线实现了高达 24% 的提升。该方法已部署到真实的 Franka Panda 机器人上,能够处理物体多种属性变化(尺寸、旋转、平移)以及布料特有属性变化(颜色、厚度、尺寸、材质)。项目概览见 https://uscresl.github.io/mail 。
英文摘要:
Some Learning from Demonstrations (LfD) methods handle small mismatches in the action spaces of the teacher and student. Here we address the case where the teacher's morphology is substantially different from that of the student. Our framework, Morphological Adaptation in Imitation Learning (MAIL), bridges this gap allowing us to train an agent from demonstrations by other agents with significantly different morphologies. MAIL learns from suboptimal demonstrations, so long as they provide $\textit{some}$ guidance towards a desired solution. We demonstrate MAIL on manipulation tasks with rigid and deformable objects including 3D cloth manipulation interacting with rigid obstacles. We train a visual control policy for a robot with one end-effector using demonstrations from a simulated agent with two end-effectors. MAIL shows up to $24\%$ improvement in a normalized performance metric over LfD and non-LfD baselines. It is deployed to a real Franka Panda robot, handles multiple variations in properties for objects (size, rotation, translation), and cloth-specific properties (color, thickness, size, material). An overview is on https://uscresl.github.io/mail .
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。