PesoBot: Precision Handling of New Philippine Peso Coins Using Vacuum Gripper and YOLOv8 Algorithm

Confusion and misidentification stemming from identical features, such as color, introduced by the New Generation Currency (NGC) Coin Series, prompted the development of PesoBot. This project provides an overview of the vacuum arm gripper, capable of identifying and distinguishing four denominations...

Full description

Saved in:
Bibliographic Details
Published in2024 7th International Conference on Informatics and Computational Sciences (ICICoS) pp. 54 - 59
Main Authors Samaniego, Ann Louiella M., Layco, Shairille P., Remot, Nicole L., Bernabe, Daniella Gracielle G., Clavecilla, Karl Patrick R., Rosales, Marife A.
Format Conference Proceeding
LanguageEnglish
Published IEEE 17.07.2024
Subjects
Accuracy
artificial intelligence
coin sorting
Grippers
PesoBot
Reliability
Robot kinematics
Safety
Training
vacuum gripper
Vacuum systems
YOLOv8
Online AccessGet full text

Cover

More Information
Summary:Confusion and misidentification stemming from identical features, such as color, introduced by the New Generation Currency (NGC) Coin Series, prompted the development of PesoBot. This project provides an overview of the vacuum arm gripper, capable of identifying and distinguishing four denominations of newly generated coins, including 1 peso, 5 pesos, 10 pesos, and 25 cents. The project used the YOLOv8 algorithm to optimize performance and functionality, enhancing the robot's ability and speed in identifying coins. These optimizations involve extensive training and testing with its dataset, adjusting the value of iteration for optimal results. The study demonstrates an impressive overall performance, achieving a remarkable mean Average Precision (mAP) of {96.46 \%} as the highest result with its performance in training 3. The results for the said denominations resulted in high accuracy and reduced chance of misclassifications. The PesoBot's ability to differentiate between coins with nearly identical appearances is highlighted with a precision of 88.16%, a recall of 89.9%, and an F1-score of 0.955. These metrics highlight the proficiency of the study in accurately detecting and recognizing various peso coins. The PesoBot object detection model ensures reliability and efficiency in real-world peso sorting tasks and prevents possible misidentification and potential financial losses.
ISSN:2767-7087
DOI:10.1109/ICICoS62600.2024.10636913