ORIENTA AI ORIENTA AI室内，还没有 GPS + 地图那样的通用空间基础设施——这，是我们要建的入口。Outdoors, GPS and Maps solved this years ago. Indoors, nothing like it exists yet — that's the gap we're closing.
85%+ 的人类时间在室内 —— 具身智能的主场。85%+ of human time is spent indoors — home turf for embodied AI.
≈99% 的生态在室外"车与路"。室内止步于封闭小图/一次性测绘。≈99% of it is built outdoors, for cars and roads. Indoors, it stops at closed, one-off surveys.
建图、对齐、标注，压进同一次采集、并行发生——快了 100 倍的动态地图。Mapping, alignment, and annotation — compressed into one parallel capture pass. A dynamic map, 100x faster.
一次采集，多层输出。One capture, multiple layers out.
全球最大单体航站楼，自研算法一次扫完，四层就绪。One of the world's largest single terminal buildings — scanned once with our own algorithm, all four layers ready.
人、机器人、智能设备——每个移动的主体既是贡献者，也是消费者。交互、导航、消费都发生在这张图。People, robots, smart devices — every moving agent is both a contributor and a consumer. Interaction, navigation, and consumption all happen on the same map.
同一套空间底座，跨三个数量级同构成立。The same spatial infrastructure, structurally identical across three orders of magnitude.
其余场景的难点，都是它的子集——架构不变，尺度不同。Every other scenario's challenges are a subset of this one — same architecture, different scale.
配送、服务、巡检、人形。每个新场地的建图标注 NRE，是规模化的分母；部署速度就是收入速度。Delivery, service, inspection, humanoid — every new site adds mapping and labeling NRE, the drag on scale. Deployment speed is revenue speed.
机场、商场、医院需要"活"的室内地图：语义、楼层、无障碍、随变更持续更新。Airports, malls, and hospitals need a "living" indoor map: semantics, floors, accessibility, continuously updated as things change.
带语义与真实坐标的空间数据，每一次经过都在生产；纯视觉路线直接受益——这张图就是定位底座与语言 grounding 层。Spatial data with semantics and real coordinates, generated on every pass. Vision-only models benefit directly — this map is both the localization base and the language-grounding layer.
具身训练要在仿真里跑亿万步——真实布局、真实语义、真实尺度，sim2real gap 从源头缩小。既是部署底座，也是训练场。Embodied training needs hundreds of millions of steps in simulation. Real layouts, real semantics, real scale — closing the sim2real gap at the source. One capture, two uses: deployment and training.
现有玩家要么专注静态扫描，要么专注单一机器人内部建图——没有人在做跨尺度、多方共享、持续更新的空间底座。Existing players either focus on static scanning or single-robot internal mapping — no one is building a cross-scale, multi-party, continuously updated spatial layer.
大尺度 + 多方共享 + 持续更新，这三个约束叠加在一起，需要重新设计整条采集-建图-标注管线——不是在现有静态扫描产品上加一个"更新"按钮就能做到的。Large scale, multi-party sharing, and continuous updates together require rebuilding the entire capture-mapping-annotation pipeline from scratch — not something you bolt onto an existing static-scan product.
单一机器人厂商的建图只服务自己的车队，不跨厂商、不跨模态共享——每换一个场景、一个厂商，地图从零开始。A single robot maker's SLAM only serves its own fleet — it doesn't share across vendors or modalities. Every new site or vendor starts the map from zero.
同一套空间底座，已经在真实环境里被不同形态的机器人验证过。The same spatial infrastructure, already validated in the real world by robots of different form factors.
核心建图定位算法完全自主知识产权，5 项发明已提交 USPTO 并缴费，锁定优先权日期。Core mapping and localization algorithm — fully self-owned IP. 5 inventions filed with USPTO, fees paid, priority date locked.
我们把空间、定位、导航与协同，装进同一个触达入口——现在，它服务着全世界最复杂的中转枢纽。We packed space, localization, navigation, and coordination into a single touchpoint — today, it serves one of the world's most complex transit hubs.
18 个月与国航的深度合作——12 个月信任建立与流程摸底 + 6 个月技术集成与内测。18-month deep partnership with Air China — 12 months earning trust and learning the workflow, 6 months of technical integration and internal testing.
旅客、航司、机场，现在加上具身智能与机器人生态——同一份地图，多条变现路径。Passengers, airlines, airports — and now the embodied AI and robotics ecosystem. One map, many revenue paths.
降低航班延误与衔接风险，提升中转保障效率，精准触达旅客。Cuts flight-delay and connection risk, sharpens transit-service efficiency, and reaches passengers with precision.
语义地图 + 实时感知，优化资源配置，提升运营效率。Semantic maps + real-time sensing optimize resource allocation and boost operational efficiency.
机器人厂商按部署接入空间数据 API，具身智能基座模型公司按训练数据 license 付费——同一张图，服务两种"非人类旅客"。Robot makers pay per deployment for spatial-data API access; embodied AI foundation model companies pay for training-data licenses — the same map serving two kinds of "non-human passengers."
做过量产自动驾驶的地图与感知栈，也做过智能机器人与端侧多模态大模型。We've shipped mapping and perception stacks for production autonomous vehicles, and built intelligent robots and on-device multimodal models.
一位来自 Waymo 的工程师即将加入我们的团队。An engineer from Waymo is joining our team soon.
Orienta 正在为具身智能构建通用、动态的室内空间基础设施，欢迎航旅生态、云与 IT 伙伴，以及具身智能 / 机器人生态的战略投资人与我们交流。Orienta is building the universal, dynamic spatial layer for embodied AI. We'd love to talk to aviation-ecosystem and cloud/IT partners, and to strategic investors across embodied AI and robotics.
预约 30 分钟通话 →Book a 30-min call →