Session Index

Optical Design

Optical Design III
Friday, Dec. 3, 2021  10:45-12:00
Presider: Prof. Pin Han Prof. Cheng-Mu Tsai
Room: 302e
10:45 - 11:00
Manuscript ID.  0089
Paper No.  2021-FRI-S0503-O001
Yu Tang Cheng Research on Design Rule of Micro-LED Backlight by Deep Reinforcement Learning
Yu Tang Cheng

In this study, we use LightTools to build a 5 x 5 Micro-LED array model with the Distributed Bragg Reflector (DBR) optical material created by RSoft. Use DBR to suppress the intensity of the active light, so that the light intensity from the side to achieve high uniformity of the light source module. Through the Reinforcement Learning (RL) to find the best solution for illuminance uniformity.

11:00 - 11:15
Manuscript ID.  0261
Paper No.  2021-FRI-S0503-O002
Yu-Cheng Sung Path Following and Obstacle Avoidance of Field Tracked Vehicles via Model Predictive Control with Deep Deterministic Policy Gradient
Yu-Cheng Sung;Shean-Jen Chen;Wen-Chuan Tseng;Chun-Ting Sung;Jen-Yu Lee

The path following and obstacle avoidance of a field tracked vehicle is conducted by using the model predict control (MPC) with the controller (agent) trained by deep deterministic policy gradient (DDPG). The MPC forces the vehicle to follow the global reference path and the DDPG agent generates a new local path to avoid obstacles.
Reinforcement learning has been successfully applied to various tasks, but still there some problems due to low sample efficiency and inability to deal with large time delay. Therefore, our work is combined the MPC with the DDPG.

11:15 - 11:30
Manuscript ID.  0483
Paper No.  2021-FRI-S0503-O003
Yong-Sheng Lin One kind of endoscope which can use to construct the three dimension model of a baglike organ
Yong-Sheng Lin;Shi-Hwa Huang;Ching-Cherng Sun

One kind of new endoscope introduced here, it consists of a new type sensor and a new optical system which can construct the three dimension model of a baglike organ.

  Preview abstract
11:30 - 11:45
Manuscript ID.  0029
Paper No.  2021-FRI-S0503-O004
Chu-En Lin Design of an optical neural network to develop a logic NAND gate with supervised learning method
Chu-En Lin;Ching-Pao Sun;Chii-Chang Chen

In this article, we establish an optical neural network to achieve a NAND gate with supervised learning method. The optimized bit error rate is 0.2495 as well as the normalized root mean square error is 0.65. Thus, through our result, our design could be a potential photonic device.

11:45 - 12:00
Manuscript ID.  0685
Paper No.  2021-FRI-S0503-O005
Ming-Hsiang Chuang Measurement of thermal stress, thermal expansion coefficient and biaxial modulus of optical notch filter
Ming-Hsiang Chuang;Chuen-Lin Tien;Hong-Yi Lin

This work is to explore the nine-layer notch filter (SiO2/Nb2O5) with the center wavelength of 480 nm deposited by electron gun evaporation technology, and using the dual substrate technology to measure the coefficient of thermal expansion (CTE), biaxial modulus and thermal stress changes of the thin film.