Generation of synthetic Force-Torque data for learning-based control of robots with Sim2Real transfer in industrial assembly processes
- Robots are widely used in industrial assembly processes for
various applications. However, conventional control
algorithms that rely on static programming are not adaptable
for facilitating human-robot collaborations and interactions
with an unknown environment. In applications such as these,
force and torque feedback become vital, and the potential of
this control strategy can be fully leveraged by training a
supervised learning algorithm with force-torque data.
However, collecting training data for learning-based force-torque control can be time-consuming and inefficient. Hence,
simulations can generate abundant data for interactions
between the robot and its environment. This paper proposes a
synthetic force-torque data generation method for hassle-free
robot training when interacting with rigid and deformable
materials in industrial assembly tasks by leveraging transfer
learning techniques. Additionally, it discusses how the
simulation model of theRobots are widely used in industrial assembly processes for
various applications. However, conventional control
algorithms that rely on static programming are not adaptable
for facilitating human-robot collaborations and interactions
with an unknown environment. In applications such as these,
force and torque feedback become vital, and the potential of
this control strategy can be fully leveraged by training a
supervised learning algorithm with force-torque data.
However, collecting training data for learning-based force-torque control can be time-consuming and inefficient. Hence,
simulations can generate abundant data for interactions
between the robot and its environment. This paper proposes a
synthetic force-torque data generation method for hassle-free
robot training when interacting with rigid and deformable
materials in industrial assembly tasks by leveraging transfer
learning techniques. Additionally, it discusses how the
simulation model of the environment can be validated to
obtain the most accurate estimates of the real forces in the
physical world with Sim2Real transformation of the industrial
assembly setup.…
Download full text files
Postprint. © 2024 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Statistics
Additional Services
