10:30 AM - 10:45 AM
▲ [19a-Z22-6] Improvement of Position Prediction for Optical Wireless Power Transmission System using Machine Learning
Keywords:Optical wireless power transmission, Machine learning, Position prediction
Wireless power transmission (WPT) is an advanced technique to transmit electric power without any cable. Optical wireless power transmission (OWPT) can transmit the electric power to longer distance (>100m) with high power density and control direction more accurate. In OWPT system using camera to recognize the target, aberration of irradiation direction due to the delay which is caused by the processing time for target recognition by the camera and PC occurs. In this case, the method to accurately irradiate the moving target by predicting its position on the next frame is needed.
Machine learning is used in OWPT system to improve the target recognition by predicting the position of the target on the next frame of captured image. In this research, artificial neural network (ANN) and long-short term memory (LSTM) methods are implemented. Using machine learning methods, position error in target recognition of OWPT system can be reduced to less than 3 pixels position error which is more than 75% improvement compared with previous experimental results.
Machine learning is used in OWPT system to improve the target recognition by predicting the position of the target on the next frame of captured image. In this research, artificial neural network (ANN) and long-short term memory (LSTM) methods are implemented. Using machine learning methods, position error in target recognition of OWPT system can be reduced to less than 3 pixels position error which is more than 75% improvement compared with previous experimental results.