Training of a robotic system have been used
around for some time. Big companies like FANUC make their
own robotic arm which are not already programmed. We have
to program it for particular task. However, we are trying to
develop a SELF EXPERIENCE replay learning technique
wherein one can train robots to perform a task by performing
the task once, MANUALLY. Such a system would reflect on
the human methods of teaching wherein a person teaches a
child how to perform a particular task by showing them how it
is done by actually performing it once himself. For the
purpose of demonstration of such a system, we will develop an
algorithm via which the robot will record the actions when
performed by the user during the ‘learning phase’ which is
nothing but when the user is performing the action for the
robot for the first time. In this paper, we explore and review
various existing technologies, techniques and work being done
on the same.
Mayuri Kanholkar : BE (Information Technology) from Priydarshanis J.L.College of
Engineering, Nagpur in 2012. Appearing ME 4th sem (Wireless
communication & computing) from G.H. Raisoni College of
Engineering & Technology for Women, Nagpur.
Hemlata Dakhore : BE (CSE) from G. H. Raisoni College of Engineering, Nagpurin
2007. MTech (CSE) from G.H. Raisoni College of Engineering,
Nagpur in 2009.
FANUC Factory automation numerical control
MANUALLY Done by any human being
SELF EXERIANCE
experience by itself
We have referred several papers, describing several
techniques to train various robotic systems based on
experience replay learning technique. Our main concern
is adopting an accurate algorithm based on experience learning approach to record action & converts them into
devised motion codes. Hence, it is important to have a
very strong algorithm. Amongst the ones that we
reviewed viz. Reinforcement Learning Technique and
Human action Replication Method , we have chosen not
to directly adopt one single algorithm but to adopt
aspects from all, thus developing a learning system
which will use to train industrial robots based on
experience replay learning technique.
[1] Sander Adam, Lucian Bus¸oniu, and Robert
Babu?ska, “Experience Replay for Real-Time
Reinforcement Learning Control", IEEE
Transactions on systems, man, and cybernetics,
march 2014.
[2] Shih Huan Tseng1, Ju-Hsuan Hua2, Shao-Po Ma3
and Li-ehen Fu4, “Human Action Replication based
Robot Performance Learning in a Social
Environment”, 2013 IEEE International Conference
on Robotics and Automation (ICRA) Karlsruhe,
Germany, May 6-10, 2013.
[3] D. Katagami, S. Yamada, “Active Teaching for an
Interactive Learning Robot”, Proceedings of IEEE
international Workshop on Robot and Human
Interactive Communication Millbrae. California.
USA. Oct. 31 -Nov. 2, 2012.
[4] JamilAbouSaleh, FakhreddineKarray, and Michael
Morckos, “A Qualitative Evaluation Criterion for
Human-Robot Interaction System in Achieving
Collective Tasks”, WCCI 2012 IEEE World
Congress on Computational IntelligenceJune, 10-
15, 2012 - Brisbane, Australia.
[5] Puran Singh*, Anil Kumar, Mahesh Vashisth,
“Design of a Robotic Arm with Gripper & End
Effector for Spot Welding”, Universal Journal of
Mechanical Engineering , 2013.
[6] Maya C. et al., "Human Preference for Robot-
Human Hand-over Configurations", IEEE/RSJ
International Conference on Intelligent Robots and
Systems, 201l.
[7] JamilAbouSaleh, FakhreddineKarray University of
Waterloo, “Towards Generalized Performance
Metrics for Human-Robot Interaction”, conference
proceedings, 2008.
[8] WolframBurgard, Armin B. Cremers, Dieter Fox,
Dirk Hahnel, Gerhard Lakemeyer , Dirk Schulz,
Walter Steiner, and Sebastian “The Interactive
Museum Tour-Guide Robot” Thrun Computer
Science Department III University of Bonn Aachen
University, Germany.