TY - GEN
T1 - Robotic writing based on trust region optimization and imitation learning
AU - Yang, Min Gyu
AU - Ahn, Kuk Hyun
AU - Song, Jae Bok
N1 - Funding Information:
This work was supported by IITP grant funded by the Korea Government MSIT. (No. 2018-0-00622)
Publisher Copyright:
© 2020 Institute of Control, Robotics, and Systems - ICROS.
PY - 2020/10/13
Y1 - 2020/10/13
N2 - As the use of robots in service area increases, it is necessary to replace human tasks in daily life with robots. Among them, this study focuses on writing and aims to write alphabets on the blackboard with chalk using a 7-DOF robot arm. In chalk writing, it is necessary to regulate the contact force measured between chalk and blackboard within appropriate value, so that the desired letters can be written consistently in the desired thickness. To this end, the deep reinforcement learning algorithm, proximal policy optimization (PPO) was used as a height regulator for the end-effector of a robot to control the contact force between the chalk and the blackboard. Moreover, imitation learning network which is trained using human handwritten data, was used as a planar path generator. The performance of the proposed height regulator and planar path generator was verified by experiments.
AB - As the use of robots in service area increases, it is necessary to replace human tasks in daily life with robots. Among them, this study focuses on writing and aims to write alphabets on the blackboard with chalk using a 7-DOF robot arm. In chalk writing, it is necessary to regulate the contact force measured between chalk and blackboard within appropriate value, so that the desired letters can be written consistently in the desired thickness. To this end, the deep reinforcement learning algorithm, proximal policy optimization (PPO) was used as a height regulator for the end-effector of a robot to control the contact force between the chalk and the blackboard. Moreover, imitation learning network which is trained using human handwritten data, was used as a planar path generator. The performance of the proposed height regulator and planar path generator was verified by experiments.
KW - Deep learning
KW - Imitation learning
KW - Reinforcement learning
KW - Robotics
UR - http://www.scopus.com/inward/record.url?scp=85098061687&partnerID=8YFLogxK
U2 - 10.23919/ICCAS50221.2020.9268403
DO - 10.23919/ICCAS50221.2020.9268403
M3 - Conference contribution
AN - SCOPUS:85098061687
T3 - International Conference on Control, Automation and Systems
SP - 132
EP - 136
BT - 2020 20th International Conference on Control, Automation and Systems, ICCAS 2020
PB - IEEE Computer Society
T2 - 20th International Conference on Control, Automation and Systems, ICCAS 2020
Y2 - 13 October 2020 through 16 October 2020
ER -