Session Index

3D Visual positioning is a critical function in applications such as navigation and extended reality (XR) experiences. Recently, machine learning technologies have been actively applied to the positioning task. In this talk, advances in machine learning-based 3D visual positioning will be firstly introduced. Also, our latest works on 3D point cloud mapping and simultaneous localization and mapping (SLAM) acceleration will be presented. By utilizing a branch convolutional neural network (B-CNN) model, the "zoom-in" equivalent property results in favorable positioning accuracy and successful real-time implementation. Furthermore, deep neural networks can also be utilized to predict the iteration outcomes from the bundle adjustment (BA) in SLAM and achieve the same level of accuracy in a much shorter period.

  Preview abstract

This study proposed an HMD based on curved holographic waveguide which consists of a cylindrical curved glass waveguide and two HOEs. A suitable coupling direction is selected to construct the cylindrical curved waveguide display. Astigmatism aberration induced by the cylindrical curved waveguide is compensated by the cylindrical lens. The FOV of this system achieve to 11.02° in the horizontal direction and 24.28° in the vertical direction.

  Preview abstract

The major challenges of augmented reality (AR) systems and virtual reality(VR) systems are varifocal images for vergence accommodation conflict (VAC) and vision corrections. We design a liquid crystal (LC) lens set consisting of three LC lenses for varifocal images and vision corrections in AR and VR. Four operating modes of such a LC lens set present three electrically tunable lens powers:0, -0.79,-2,and-3.06 diopters. We also demonstrate AR and VR systems by adopting the LC lens set to exhibit functions of varifocal images and vision corrections which enable to solve VAC as well as vision problem in AR and VR.

  Preview abstract

We propose a novel method of single-shot fringe projection incoherent digital holography to perform three-dimensional (3-D) diffused face detection. Preliminary results show that the combination of incoherent digital holography and fringe projection profilometry can perform wide field of view and fast 3-D surface imaging of a simulating facial model under incoherent or white light source illumination.

  Preview abstract

We have described the design process of the diffractive waveguide for augmented reality display, capturing the link between the ray-tracing result and the choice of the grating structure. Our design process uses computation-guided rules derived from ray-tracing and diffraction theory, which produces designs that fulfill critical requirements such as field of view, high diffraction efficiency, image uniformity, and a compact layout. Practically, we have designed a display with over 40 degrees field of view and with a 12 by 10 mm eye box. Furthermore, the design process is robust and applicable to most AR design architectures.

  Preview abstract

In this paper, we demonstrate design and operation principle of holographic waveguide for augmented reality by using volume holographic optical element (VHOEs). The optical experiments have been performed to demonstrate our recursive inference. Then, we propose a novel optical scheme to increase the field of view of a full-color VHOE-based waveguide with 2D eye-box expansion.

  Preview abstract

We apply light field technique for our mixed reality (MR) glasses design. For light field display, resolution loss is a major issue we need to conquer. In this study, preliminary experiment for light field design is presented, and an important condition for light field resolution analysis is also proposed.

  Preview abstract

We propose two kinds of polarizer-free and electrically controllable liquid crystal (LC) ambient-light modulator with different dynamic tunable ranges, 6.86%-37.5% and 0.27% and 12.37%, in visible wavelength. The key mechanism is the guest-host effect of dichroic dyes and LCs. The former can be a dimmer for AR glasses and the latter a modulator for switching glasses between AR and VR.

  Preview abstract

We propose a hybrid approach combining the rigorous coupled wave analysis (RCWA) and the Monte Carlo ray-tracing method to design augmented reality (AR) devices. Any situation with both geometric rays and diffracted waves conditions can quickly obtain theoretical results to reduce the cost of trial-and-error and increase research efficiency.

  Preview abstract

In this study, the Volume Hologram being an Interference of the Lights emitted from elementary Light Sources (VOHIL) is used to build the numerical model of the volume holographic optical element, which can accurately predict the trade-off between the wavelength selectivity and the angular selectivity. And we also validate each other with the experimental results and the simulation results.

  Preview abstract

We propose a brand new eye tracking technique that is applied to the light-field based near-eye display. The convolution neural network (CNN) is used to predict the glaze point of eye (x-direction pixel, y-direction pixel). A formula is proposed to calculate the depth of the eye (d_e). By exploiting de, we can produce more precise light field image of MR device. Moreover, the information of the glaze point can provide optional operation method of MR device.

  Preview abstract

PointNet2 can classify and semantically segment point cloud. We will use PointNet2 to learn the matrix space for point cloud. Learning how machine learning works and focus on segmentation for object.Finally, divide the object into different parts and mark them.

  Preview abstract

We investigate resolution enhancement technology of a near infrared metalens. First, we make a near-infrared light metalens, then using a simple and efficient fabrication by trimming PR method with oxygen plasma. Finally, photoresist trimming for resolution enhancement technology could increase the focusing efficiency from 6.1% to 45.1%, and also push the resolution limit from 350 nm to 260 nm on i-line stepper.

  Preview abstract

It has become possible to manipulate light spectrally and spatially on demand at wavelength scale with the development of metasurfaces in recent years. However, large-scale penetration of achromatic metasurface-based metalens are current challenges. Here, I report our recent developments of dispersion-engineered metasurfaces and centimeter-scale diameter RGB-achromatic metalens by using inverse-design method. We demonstrate virtual reality and augmented reality platforms by using our RGB-achromatic metalens with a fiber scanning near-eye display or a laser-illuminated micro liquid crystal display. Such large-scale meta-optics opens a new paradigm for applications and inspire a vision for a future in which meta-optics will be widely used.

  Preview abstract

This work proposes a brand new approaching for the laser ablation study of Si3N4 film based upon unsupervised learning in CMOS-MEMS process. The study demonstrates the interval time is the dominant role on laser ablation quality. There are four classes for this task by k-means algorithm. While the interval time is larger than 60s, the mean of laser ablation quailty(reb) is more than 82%. The interval time is lower than 40s which reb is less than 71%. The result shows energy 0.318mJ, pulse shots 3 times, and pad position right gets the more reb compared to other conditions.

  Preview abstract

Diffractive waveguide display, which has shown promising results in the field of Metaverse, provides thin and light optical solutions for augmented reality devices. While the design of diffractive combiners, which requires the designer to consider both optically and mechanically, is a challenging task for engineers who barely learn the knowledge of diffractive optics, thus complicating the design process and slowing down the iterative development. Based on the design principle of diffractive optics, we have developed a handy online application to design the layout of diffractive combiners with pupil expansions.

  Preview abstract

In metaverse photonics, all the contents might be presented in a micro display for further projection. This work explores the calibration process on the brightness uniformity for the micro-display. Here, the calibration is extensive down to pixel-wise such that all the image contents could be reprojected onto human retina with high quality for metaverse applications.

  Preview abstract

In this paper, we present the feasibility of compensation of color breaking in a holographic-optical-element based augmented reality (AR) display. We first build up a theoretic optical model of the AR display system to evaluate the color breaking of displayed white image from the user’s point of view. During the calculation, the color coordinate of each pixel of projected virtual image can be obtained. Therefore, based on those calculated results, we propose a color correction algorithm by applying a pre-compensation scheme to the R/G/B values of each display pixel. This method enables the full-color image in AR system applications.

  Preview abstract

The vision correction for near-eye display by using an electrically tunable lens based on the spatially extended LC lens is demonstrated in this paper. For the first time, we found that the spatially extended LC lens shows the spatially distribution in lens power ranging from +3.25D to -3.25D. The surprising results give LC lenses an insightful aspect and overrule the traditional statement of the limitation of lens power set by the optical phase difference in gradient-index LC lenses. It also paves a way in ophthalmic applications.

  Preview abstract

In this study, the proposed near-eye display (NED) system is composed of two holographic waveguides with wavelength multiplexing. Each holographic waveguide is designed to offer half FOV for the whole system. After combination of these two waveguides, rhe NED system FOV can reach 50°×20.71° (H×V).

  Preview abstract

A Maxwellian-view display based on HOE with large FOV is proposed. The HMD system can provide information from mask and image will not defocus when observer focus on different distance. The FOV of the demonstration system can reach 39.38 °×56.83° (H×V).

  Preview abstract

In this study, the research topic is the electrically tunable focusing large aperture liquid crystal lens with three-zone and three-dimensional curved electrodes. The three-dimensional curved electrodes are designed by separating the ideal phase profile into three zones according to the structures of Fresnel lenses, and the optical path difference between any two adjacent zones of the three-dimensional electrode is an integer multiple of 2π. The performances of the LC lens, such as the electrically tunable focus, response time, and others, are also discussed.

  Preview abstract

We demonstrate a single-shot and image-based polarization detection system for material recognition. The stokes parameters are measured under a single-shot measurement using 4 electrically tunable liquid crystal wave plates, 4 polarizers and 4 camera modules. The optical principle is introduced and the experiments are performed. We also use a metallic plate and a glass substrate to demonstrate the material recognition. The impact of this study is to provide a practical way in image-based polarization detection in Advanced Driver Assistance Systems for material recognition which could help in driving safety.

  Preview abstract

In this research, we demonstrated a red-orange organic light emitting diode (OLED) through solution
process by incorporating a novel synthesized thermally activated delayed fluorescence (TADF) emitter. By
optimizing the thermal annealing duration of emitting layer, the maximum current efficiency of 0.83 cd/A, emission
peak wavelength of 590 nm and CIE coordinates of (0.49, 0.45) can be achieved.

  Preview abstract

In this research, a novel green thermally activated delayed fluorescence organic lighting-emitting diode(TADF) material based on Carbazole-derivative was synthesized and employed as high efficiency organic light-emitting diode (OLED). By using the 4CzIPN doped in the new green TADF host, maximum external quantum efficiency (EQE) reached 10.44% with CIE coordinates of (0.2792, 0.5784).

  Preview abstract

In this research, a new triplet-triplet annihilation (TTA) emitter based on pyrene derivative was synthesized and employed as the host material for high efficiency blue organic light-emitting diode (OLED). By optimizing the dopant concentration of fluorescent emitter in the emitting layer (EML), the maximum external quantum efficiency (EQEmax) of 6.93% with CIE coordinates of (0.136, 0.192) can be achieved. The time-resolved electroluminescence measurement revealed that the highest delayed emission ratio of 20% contributed to device efficiency.

  Preview abstract

In our research, we design three gratings to realize the AR glasses at wavelength 520nm. Besides that, we optimize the structure of grating to obtain uniform output efficiency. However, slant grating is difficult to fabricate through photolithography. To solve this problem, metagrating is a way to deal with it. By using double patterning technique, we succesfully build a metagrating which critical dimension reaches 50nm via KrF scanner.

  Preview abstract

The aim of this study was to construct a high quality and compact floating projection system using several free-form reflectors to achieve the floating effect and correct for aberrations. The system is 13cm x 12cm x 14cm and can float an image of approximately 8cm. The final evaluation of the image quality in terms of MTF and aberration has proven that the system can achieve a high-quality floating display.

  Preview abstract

Metalenses are a promising new optical technology with applications in AR/VR, biomedical imaging, and wearable consumer electronics. Owing to their planar profiles and versatility, they allow for a dramatic miniaturization of existing optical systems while realizing complex new optical functions. MetaOptic Designer, a fully automated design tool has been developed using inverse design techniques adapted for systems containing cascaded metasurfaces with arbitrary configurations of parameterized meta-atoms. The optimized layout is obtained automatically based on specified target functions. The performance of the optimized metalens system can then be validated by different simulation approaches. Several design examples will be presented to demonstrate the capability and usability of this powerful design flow.

  Preview abstract

Comparison from technical to performance for different optical configuration, micro display panel and optical combiner will be given during the talks, which focus on AR optical engine to glasses especially. Several AR HMD and Glasses will be illustrated as example, and newest process in industry will be included also.

  Preview abstract

Bundle adjustment (BA) is the major optimization step which simultaneously refines 3D coordinates and accounts for a large portion of execution time in visual simultaneous localization and mapping (SLAM). Recent solutions adopting iterative and originally slow Gaussian belief propagation (GBP) show its potential to be fast and accurate on emerging computation platforms. We propose a novel GBP-Learning architecture with deep neural networks to avoid the dependency of an arbitrary damping factor for GBP to be stabilized. Moreover, compared with standard GBP, the learning-based approach achieves the same level of accuracy while running 17.7 times faster under GPU acceleration.

  Preview abstract

A client-server architecture for mixed-reality communication between the real world and a virtual object is proposed. We demonstrate the framework combined with text spotting model on HoloLens 2 so that the devices can receive the surrounding information to improve user experiments. Finally, our experiment results show that the client-server framework can provide on average 0.77 seconds of computational time per frame, which is not only 11.8 times faster than the client-only framework but also achieves near real-time computation. Moreover, the hardware usage of the client-side can be reduced to half of the client-only framework.

  Preview abstract