TY - GEN
T1 - Machining Learning for 2D-SLAM Object Classification and Recognition
AU - Yu, Chun Yen
AU - Peng, Chao Chung
N1 - Funding Information:
ACKNOWLEDGMENT This work was supported by the Ministry of Science and Technology under Grant No. MOST 107-2221-E-006-114-MY3 and MOST 108-2923-E-006-005-MY3.
Publisher Copyright:
© 2020 IEEE.
PY - 2020/9/28
Y1 - 2020/9/28
N2 - Artificial intelligence has being widely used in the last decade, and the relevant machine learning (ML) topics are also attracted more and more attention. Therefore, in this note, certain ML related algorithms are integrated into two dimensional simultaneous localization and mapping (2D-SLAM) technology in order to give meaningful labels to certain specific objects in the environment. For the 2D-SLAM technology, the main objective is to reconstruct a binary map of the unknown environment. However, 2D-SLAM itself does not have the capability of recognizing scanning objects. To extend the application of 2D-SLAM, in this study, point cloud clustering in conjunction with image preprocessing are used in the ML methods. The proposed method can predict the label of the clustered point clouds. Consequently, the 2D-SLAM could achieve environment awareness applications, for example, forest tree counting use. Based on a given bunch of training data, this research show that the model training accuracy is 99.89%, the validation accuracy is 97.96%, and the testing accuracy could achieve 95.96%; Finally, the predicted accuracy for the given SLAM map can be up to 80%.
AB - Artificial intelligence has being widely used in the last decade, and the relevant machine learning (ML) topics are also attracted more and more attention. Therefore, in this note, certain ML related algorithms are integrated into two dimensional simultaneous localization and mapping (2D-SLAM) technology in order to give meaningful labels to certain specific objects in the environment. For the 2D-SLAM technology, the main objective is to reconstruct a binary map of the unknown environment. However, 2D-SLAM itself does not have the capability of recognizing scanning objects. To extend the application of 2D-SLAM, in this study, point cloud clustering in conjunction with image preprocessing are used in the ML methods. The proposed method can predict the label of the clustered point clouds. Consequently, the 2D-SLAM could achieve environment awareness applications, for example, forest tree counting use. Based on a given bunch of training data, this research show that the model training accuracy is 99.89%, the validation accuracy is 97.96%, and the testing accuracy could achieve 95.96%; Finally, the predicted accuracy for the given SLAM map can be up to 80%.
UR - http://www.scopus.com/inward/record.url?scp=85098465706&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85098465706&partnerID=8YFLogxK
U2 - 10.1109/ICCE-Taiwan49838.2020.9258112
DO - 10.1109/ICCE-Taiwan49838.2020.9258112
M3 - Conference contribution
AN - SCOPUS:85098465706
T3 - 2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020
BT - 2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 7th IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020
Y2 - 28 September 2020 through 30 September 2020
ER -