학술논문
Robotic Table Tennis: A Case Study into a High Speed Learning System
Document Type
Working Paper
Author
D'Ambrosio, David B.; Abelian, Jonathan; Abeyruwan, Saminda; Ahn, Michael; Bewley, Alex; Boyd, Justin; Choromanski, Krzysztof; Cortes, Omar; Coumans, Erwin; Ding, Tianli; Gao, Wenbo; Graesser, Laura; Iscen, Atil; Jaitly, Navdeep; Jain, Deepali; Kangaspunta, Juhana; Kataoka, Satoshi; Kouretas, Gus; Kuang, Yuheng; Lazic, Nevena; Lynch, Corey; Mahjourian, Reza; Moore, Sherry Q.; Nguyen, Thinh; Oslund, Ken; Reed, Barney J; Reymann, Krista; Sanketi, Pannag R.; Shankar, Anish; Sermanet, Pierre; Sindhwani, Vikas; Singh, Avi; Vanhoucke, Vincent; Vesom, Grace; Xu, Peng
Source
Subject
Language
Abstract
We present a deep-dive into a real-world robotic learning system that, in previous work, was shown to be capable of hundreds of table tennis rallies with a human and has the ability to precisely return the ball to desired targets. This system puts together a highly optimized perception subsystem, a high-speed low-latency robot controller, a simulation paradigm that can prevent damage in the real world and also train policies for zero-shot transfer, and automated real world environment resets that enable autonomous training and evaluation on physical robots. We complement a complete system description, including numerous design decisions that are typically not widely disseminated, with a collection of studies that clarify the importance of mitigating various sources of latency, accounting for training and deployment distribution shifts, robustness of the perception system, sensitivity to policy hyper-parameters, and choice of action space. A video demonstrating the components of the system and details of experimental results can be found at https://youtu.be/uFcnWjB42I0.
Comment: Published and presented at Robotics: Science and Systems (RSS2023)
Comment: Published and presented at Robotics: Science and Systems (RSS2023)