An Automated Assembly of 3D Point Clouds using Coupling Matching and Path Planning Algorithm by Reinforcement Learning

DOI

10.2991/aer.k.201221.061How to use a DOI?

Keywords

3D Point Cloud, Assembly Path Planning, Coupling Matching, Object Assembly, Point Cloud Registration, Reinforcement Learning

Abstract

3D point cloud techniques have been playing critical roles in industrial automation. Applications focused on automated assembly are thus becoming important in manufacturing industry to solve some particular problems. To implement automated assembly, Reinforcement Learning is employed in for planning the optimized assembly path. The process structure is separated into training stage and testing stage. In the training stage, one of the objects is matched to an assembled model by using registration methods, RANdom SAmple Consensus and Iterative Closest Point, to determine the transformation between the two point clouds. To realize planning assembly path, the training is based on Q-learning method by reinforcement learning. In the testing stage, the optimized assembly path can be computed from the Q table obtained by training stage. The tested objects are three 3D point cloud data with the coupling matching part. The sockets which are female and male part are identified to match each couple part. Every object is firstly trained for every zone and next for testing. The working environment is a fixed with x, y, z coordinate which is converted to the position. The first object is trained and tested for all zones. The complete optimized assembly path planning task has been successfully achieved for representative objects. The feasibility and effectiveness of the proposed approaches have been validated by experiments.

Copyright

© 2020, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Dianthika Puteri Andini
AU  - Muhammad Yusuf Fadhlan
PY  - 2020
DA  - 2020/12/22
TI  - An Automated Assembly of 3D Point Clouds using Coupling Matching and Path Planning Algorithm by Reinforcement Learning
BT  - Proceedings of the International Seminar of Science and Applied Technology (ISSAT 2020)
PB  - Atlantis Press
SP  - 371
EP  - 378
SN  - 2352-5401
UR  - https://doi.org/10.2991/aer.k.201221.061
DO  - 10.2991/aer.k.201221.061
ID  - Andini2020
ER  -