Abstract
Besides direct interaction, human hands are also skilled at using tools to manipulate objects for typical life and work tasks. This paper proposes DeepClaw 2.0 as a low-cost, open-sourced data collection platform for learning human manipulation. We use an RGB-D camera to visually track the motion and deformation of a pair of soft finger networks on a modified kitchen tong operated by human teachers. These fingers can be easily integrated with robotic grippers to bridge the structural mismatch between humans and robots during learning. The deformation of soft finger networks, which reveals tactile information in contact-rich manipulation, is captured passively. We collected a comprehensive sample dataset involving five human demonstrators in ten manipulation tasks with five trials per task. As a low-cost, open-sourced platform, we also developed an intuitive interface that converts the raw sensor data into state-action data for imitation learning problems. For learning-by-demonstration problems, we further demonstrated our dataset’s potential by using real robotic hardware to collect joint actuation data or using a simulated environment when limited access to the hardware.
Highlights
Learning from human behaviors is of great interest in robotics (Argall et al, 2009; Kroemer et al, 2019; Osa et al, 2018)
We propose DeepClaw 2.0 (Figure 1) as a data collection platform for imitation learning, where a human teacher operates a pair of modified kitchen tongs to perform object manipulation
Our results demonstrate the potential to collect dexterous manipulation data from such configuration, which can be translated for imitation learning at a low cost in data collection
Summary
Learning from human behaviors is of great interest in robotics (Argall et al, 2009; Kroemer et al, 2019; Osa et al, 2018). Dexterous operation of various tools plays a significant role in the evolution of human behaviors from ancient times (Kaplan et al, 2000) to modern civilization (Brown and Sammut, 2011). For imitation-based manipulation learning, it is common to collect behavior cloning data by directly observing the human hand (Christen et al, 2019) or through human-guided robot demonstration (Chu et al, 2016). It is widely recognized that such dexterity in manipulation is tightly related to the sense of touch through the fingers (Billard and Kragic, 2019), challenging to model and reproduce with current development in low-cost sensing solutions. A lack of low-cost, efficient, shareable, and reproducible access to the manipulation data remains a challenge ahead
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.