空间计算行业图谱

（2024年第6期总第38期） 2024年4月12日 空间计算行业图谱1 科创金融研究中心 朱雅姝，安砾 【摘要】空间计算是在利用空间的特有属性来指导计算过程，在三维的基础之上，以人为本，连通及融合虚拟世界与现实世界的一个全新的计算范式。本文首先从技术定义出发，阐明空间计算的概念和发展历程，介绍空间计算全球的政策情况和应用场景，重点从技术流程和产业链两个角度对空间计算行业进行分析，并进一步对全球市场现状和研发进度做了深入的调研定位。 空间计算和数字孪生都是构建元宇宙的基础技术，以创造出更加逼真、开放和互动的虚拟世界。下面简要介绍其区别与联系，关于数字孪生的行业分析详见《数字孪生行业图谱》2。空间计算是一种计算范式，它利用空间感知和环境感知技术，将数字信息与物理空间相 结合，创造出一种与现实世界更加融合的交互体验。而数字孪生则是一种虚拟模型，它通过收集、整合现实世界物体或系统的数据，实时反映其状态、行为和性能，并且能够模拟其未来发展和行为。空间计算提供了将数字信息与物理空间相结合的能力，而数字孪生则提供了模拟和管理物理系统的能力。 空间计算的应用场景可涵盖零售业、远程教育、广告、游戏、医疗、军事、演示、培训、旅游等，为用户提供更加沉浸式、交互性强，并且与现实世界紧密结合的计算体验。空间计算也在仓库自动化、自动驾驶汽车和供应链自动化中发挥作用。 从技术角度，空间计算技术链上游为空间数据采集。空间数据可以分为场景数据、网格数据和深度数据，这些数据的采集基于结构光传感器、飞行时间传感器等。中游包括空间数据处理、空间数据显示，空间数据处理涉及到空间计算芯片和机器学习、数据处理算法，以实现空间定位、运动跟踪、手势交互等功能。空间数据显示在硬件上主要分为光学器件和显示屏，在算法上包括场景渲染、环境映射、虚实整合等。下游为空间数据交互，硬件上为终端，软件上包括操作系统、开发工具和内容应用部分。 从产业角度，产业链核心技术集中在上中游，硬件方面包括传感器、芯片和光学器件的制造，算法方面包括动态环境建模、实时3D图形生成、空间定位等技术。中国的优势在于光学器件和显示屏制造、软件开发与内容创作。 从行业角度，空间计算国际头部企业较多，国际企业技术平台主要有Microsoft、Meta、Magic Leap、Google、Apple、Nreal等，国 内头部企业较少，如PICO、维享时空、宸境科技、虚拟动点等。但国内在空间计算产业链一些环节中发挥重要作用，如传感器方面：韦尔股份、水木智芯；光学器件方面：蓝特光学、水晶光电、韦尔股份等；显示屏方面：利亚德、奥拓电子、中微公司、精测电子等；芯片方面：万有引力、之行无界、瑞芯微等；整机代工方面：歌尔股份、闻泰科技、兆威机电、长盈精密等；开发工具方面：华为、爱奇艺等；内容方面：腾讯、网易、风语筑等。随着元宇宙的兴起，以及Meta在消费级空间计算上的商业成功，中国的空间计算市场迎来了第二次风口，华为、小米和OPPO等纷纷推出空间计算产品。 从科研角度，空间计算领域中国研究者发力点多集中在产业链中的某一环节，而非全流程都涉及，主要集中在光学硬件、视觉算法和内容端。 ResearchReport Atlas ofthe Spatial Computing Industry3 Research Center for Sci-Tech and Finance YashuZhu, Li An Abstract: Spatial computing is a new computing paradigm that utilizes the uniqueproperties of space to guide thecalculation process, based on three-dimensional principles, with a people-oriented approach, connecting andintegrating the virtual world and the real world. In the report, we first give abriefintroduction about spatial computing,including the technologydefinition,the development process,the global policy situation andapplication. Then we analyze the spatial computing industry from theperspectivesof technical processes and industrial chains, and furtherconducts in-depth research and positioning onthe global market status andresearch progress. Spacecomputing and digital twins are fundamental technologies for building the metaverse to create a more realistic, open, and interactive virtualworld. Here is their differences and connections. For industry analysis ofdigital twins, please refer to the "Atlas of the Digital Twin Industry"4.Spacial computing isa computing paradigm that utilizes spatial andenvironmental perception technologies to combine digital information withphysical space, creatingan interactive experience that is more integratedwith the real world. However, digital twin is a virtual model that collects andintegrates data from real-world objects or systems to reflect their state,behavior and performance in real-time, which can simulate their futuredevelopment and behavior. Spacial computing provides the ability tocombine digital information with physical space, while digital twins providethe ability to simulate and manage physical systems. The application scenarios of spatialcomputing include retail, remoteeducation, advertising, games, healthcare, military, demonstration, training,tourism, etc., providing users with a more immersive, interactive, andcloselyintegrated computing experience with the real world.Spatialcomputing also plays a role in warehouse automation, autonomous vehicleand supply chain automation. Fromtechnical perspective,the upstream of the spatial computingtechnology chain is the collection of spatial data. Spacial data can be dividedinto scene data, grid data, and depth data, which are collected based onstructured light sensors, Time of flight sensors, etc. The midstream includes spatial data processing, spatial data display. Spatial data processing involvesspatial computing chips, machine learning, and data processing algorithmsto achieve functions such as spatial positioning, motion tracking and gestureinteraction. Spacial data display is mainly divided into optical devices anddisplay screens in hardware, and algorithms include scene rendering,environment mapping, virtual and real integration, etc. Downstream isinteraction using spatial data, hardware is the terminal, and softwareincludes operating systems, development tools and content application parts.From industrial perspective, the core technologies are concentrated in theupper and middle of the spacial computing industrial chain. In terms ofhardware,including the manufacturing of sensors,chips,and opticalcomponents. In terms of algorithms, including technologies such as dynamicenvironmentmodeling,real-time 3D graphics generation,and spatialpositioning.China's advantages lie in the manufacturing of opticalcomponentsand display screens,software development,and contentcreation. From sector perspective, there are many leading international companies inspatialcomputing,with international enterprise technology platformsincluding Microsoft, Meta, Magic Leap, Google, Apple, Nreal, etc. Thereare rel