Physical Review B

We study the problem of efficient semantic segmentation for large-scale 3D point clouds. By relying on expensive sampling techniques or computationally heavy pre/post-processing steps, most existing approaches are only able to be trained and operate over small-scale point clouds. In this paper, we introduce RandLA-Net, an efficient and lightweight neural architecture to directly infer per-point semantics for large-scale point clouds. The key to our approach is to use random point sampling instead of more complex point selection approaches. Although remarkably computation and memory efficient, random sampling can discard key features by chance. To overcome this, we introduce a novel local feature aggregation module to progressively increase the receptive field for each 3D point, thereby effectively preserving geometric details. Extensive experiments show that our RandLA-Net can process 1 million points in a single pass with up to 200x faster than existing approaches. Moreover, our RandLA-Net clearly surpasses state-of-the-art approaches for semantic segmentation on two large-scale benchmarks Semantic3D and SemanticKITTI.

/pdf/randla-net-efficient-semantic-segmentation-of-large-scale-4msnszot0z.pdf

RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds

Thermal radiation is one of the most universal physical phenomena, and its study has played a key role in the history of modern physics. Our understanding of this subject has been traditionally bas...

Radiative Heat Transfer

Hyperbolic metamaterials and their applications

United States. Advanced Research Projects Agency-Energy (Awards DE-AR0000471 and DE-AR0000181)

Concentrating Solar Power.

An omnidirectional nanophotonic solar selective absorber is fabricated on a large scale using a template-stripping method. The nanopyramid nickel structure achieves an average absorptance of 95% at a wavelength range below 1.3 μm and a low emittance less than 10% at wavelength >2.5 μm.

Large-Scale Nanophotonic Solar Selective Absorbers for High-Efficiency Solar Thermal Energy Conversion.

In the near field, radiative heat transfer can exceed the prediction from Planck's law by several orders of magnitude, when the interacting materials support surface polaritons in the infrared range. However, if the emitter and absorber are made from two different materials, which support surface polariton resonances at different frequencies, the mismatch between surface polariton resonance frequencies will drastically reduce near-field radiative heat transfer. Here, we present a broadband near-field thermal emitter/absorber based on hyperbolic metamaterials, which can significantly enhance near-field radiative heat transfer with infrared surface-polariton-resonance materials and maintain the monochromatic characteristic of heat transfer. Instead of using an effective medium approximation, we perform a direct numerical simulation to accurately investigate the heat transfer mechanisms of metamaterials based on the Wiener chaos expansion method.

https://link.aps.org/accepted/10.1103/PhysRevB.87.115403

Broadband near-field radiative thermal emitter/absorber based on hyperbolic metamaterials: Direct numerical simulation by the Wiener chaos expansion method

We present a review of 3D point cloud processing and learning for autonomous driving. As one of the most important sensors in autonomous vehicles (AVs), lidar sensors collect 3D point clouds that precisely record the external surfaces of objects and scenes. The tools for 3D point cloud processing and learning are critical to the map creation, localization, and perception modules in an AV. Although much attention has been paid to data collected from cameras, such as images and videos, an increasing number of researchers have recognized the importance and significance of lidar in autonomous driving and have proposed processing and learning algorithms that exploit 3D point clouds. We review the recent progress in this research area and summarize what has been tried and what is needed for practical and safe AVs. We also offer perspectives on open issues that are needed to be solved in the future.

3D Point Cloud Processing and Learning for Autonomous Driving: Impacting Map Creation, Localization, and Perception

Although blackbody radiation described by Planck's law is commonly regarded as the maximum of thermal radiation, thermal energy transfer in the near-field can exceed the blackbody limit due to the contribution from evanescent waves. Here, we demonstrate experimentally a broadband thermal energy extraction device based on hyperbolic metamaterials that can significantly enhance near-field thermal energy transfer. The thermal extractor made from hyperbolic metamaterials does not absorb or emit any radiation but serves as a transparent pipe guiding the radiative energy from the emitter. At the same gap between an emitter and an absorber, we observe that near-field thermal energy transfer with thermal extraction can be enhanced by around 1 order of magnitude, compared to the case without thermal extraction. The novel thermal extraction scheme has important practical implications in a variety of technologies, e.g., thermophotovoltaic energy conversion, radiative cooling, thermal infrared imaging, and heat assisted magnetic recording.

Near-field energy extraction with hyperbolic metamaterials.

Thermal radiation with a narrow-band emission spectrum is of great importance in a variety of applications such as infrared sensing, thermophotovoltaics, radiation cooling, and thermal circuits. Although resonant nanophotonic structures such as metamaterials and nanocavities have been demonstrated to achieve the narrow-band thermal emission, maximizing their radiation power toward perfect emission still remains challenging. Here, based on the recently developed quasi-normal mode theory, we prove that thermal emission from a nanoscale transmission line resonator can always be maximized by tuning the waveguiding loss of the resonator or bending the structure. By use of nanoscale transmission line resonators as basic building blocks, we experimentally demonstrate a new type of macroscopic perfect and tunable thermal emitters. Our experimental demonstration in conjunction with the general theoretical framework from the quasi-normal mode theory lays the foundation for designing tunable narrow-band thermal emit...

Baoan Liu

Papers

Large-Scale Nanophotonic Solar Selective Absorbers for High-Efficiency Solar Thermal Energy Conversion.

Broadband near-field radiative thermal emitter/absorber based on hyperbolic metamaterials: Direct numerical simulation by the Wiener chaos expansion method

3D Point Cloud Processing and Learning for Autonomous Driving: Impacting Map Creation, Localization, and Perception

Near-field energy extraction with hyperbolic metamaterials.

Perfect Thermal Emission by Nanoscale Transmission Line Resonators