机器人现场做咖啡、画糖画！跨维智能数博会展位亮点抢先看Robots make coffee and paint candy on site! Get a sneak peek at the highlights of the booth at the Cross dimensional Intelligent Data Expo

贵阳网·贵阳新闻讯 当机器人不仅能握手、跳舞，还能现场为你做一杯咖啡、画一幅糖画，具身智能才算真正走入了人间烟火。

2026中国国际大数据产业博览会即将在贵阳召开。作为深圳本土培育的具身智能独角兽企业，跨维（深圳）智能数字科技有限公司将亮相E1馆E1F01展位，带来国内首款“AI引擎-运动控制-大脑-本体”全链路自研的第二代通用人形机器人DexForce W1 Pro。届时，观众不仅能近距离观察机器人如何制作冰淇淋、爆米花，还能现场体验非遗文化具身智能制作糖人的场景体验。

跨维智能的展位设计以新中式国风美学为基底，融合人工智能的科技属性。据企业对接人陈泓介绍，展位现场将设置糖画、咖啡、冰淇淋、爆米花四大业态同步呈现，“观众从下单到付款再到机器人递送，可以体验全流程标准化服务。糖画机器人还能复刻观众三笔以内的简笔画创意，做出一份可食用的‘AI纪念品’。”

这并非简单的“炫技”。以咖啡制作为例，DexForce W1 Pro需完成识别抓取咖啡胶囊、精准放入咖啡机、操作按钮、接取冲泡咖啡等一系列连贯动作，对机器人的环境感知精度、动作控制精度和长任务规划能力都是硬核考验。

“我们想打破行业里那种‘概念感’。” 陈泓表示，“让观众亲眼看到、亲口尝到，具身智能不是实验室原型，而是已经走进日常商业场景的成熟技术。”

如果说现场演示展现的是单点技能，那么规模化商业落地才是检验技术价值的真正标尺。陈泓透露，DexForce W1 Pro目前已在全国超过15个城市的多类型商业场景实现常态化运营，覆盖智能交互迎宾、商品服务、便民运营等多元业态。

这一落地能力背后，是跨维智能构建的“数据-模型-本体-场景”全链条闭环。2026年6月，跨维发布DexBEV具身统一空间表征框架，将自动驾驶领域的BEV鸟瞰范式引入机器人领域；同期推出的Aligned DexWorld成为全球首个贯通“人类-仿真-真机”的时空动作对齐大规模数据集。而在WAIC 2026上首次公开亮相的自研仿真引擎DexVerse™，更是全球唯一支持同场景下多物理场耦合的具身智能仿真引擎。这套技术基座让跨维在真实产业场景的落地中具备了持续迭代的“造血能力”。

在陈泓看来，当前具身智能行业百花齐放，但真正意义上的商业化规模还没有完全跑通。“跨维的思路是‘沿途下蛋’——把50多个真实落地场景当成通用算法进化的‘培养皿’，以复杂实操任务反哺大脑泛化能力。”陈泓说，“数博会是一个很好的窗口，我们希望在贵阳这个算力枢纽城市，让更多人看到具身智能不只是一个遥远的概念——而是已经逐渐从实验室开始商业化落地了。回顾过去两年前的具身智能磕磕绊绊，但现在已经在各行各业开始出现密集型的应用，模型离不开算力，贵阳作为中国算力节点城市，基础非常好，也非常适合这一产业。

记者 刘云嵩

Guiyang News: Being a robot can not only shake hands and dance, but also make a cup of coffee and draw a sugar painting for you on site. Only with embodied intelligence can you truly step into the world of fireworks.

The 2026 China International Big Data Industry Expo is about to be held in Guiyang. As a locally cultivated embodied intelligent unicorn enterprise in Shenzhen, Cross Dimensional (Shenzhen) Intelligent Digital Technology Co., Ltd. will debut at booth E1F01 in Hall E1, bringing the first domestically developed second-generation universal humanoid robot DexForce W1 Pro with a full chain of "AI engine motion control brain ontology". At that time, the audience will not only be able to observe up close how robots make ice cream and popcorn, but also experience the scene of intangible cultural heritage embodied intelligence making sugar figurines on site.

The booth design of Cross dimensional Intelligence is based on the new Chinese style aesthetics and integrates the technological attributes of artificial intelligence. According to Chen Hong, the contact person of the enterprise, four major formats of candy painting, coffee, ice cream, and popcorn will be presented simultaneously at the booth. "Visitors can experience standardized services throughout the entire process from placing orders to payment to robot delivery. The candy painting robot can also replicate the audience's simple strokes within three strokes, making an edible 'AI souvenir'. ”

This is not just a simple 'show off'. Taking coffee making as an example, DexForce W1 Pro needs to complete a series of coherent actions such as recognizing and grabbing coffee capsules, accurately placing them into the coffee machine, operating buttons, and picking up and brewing coffee, which are hardcore tests for the robot's environmental perception accuracy, action control accuracy, and long-term task planning ability.

We want to break the 'conceptual sense' in the industry, "Chen Hong said." Let the audience see and taste firsthand that embodied intelligence is not a laboratory prototype, but a mature technology that has already entered daily commercial scenarios

If on-site demonstrations showcase single point skills, then large-scale commercial implementation is the true yardstick for testing the value of technology. Chen Hong revealed that DexForce W1 Pro has now achieved normalized operation in various commercial scenarios in more than 15 cities across the country, covering diverse formats such as intelligent interactive welcome, product services, and convenient operations.

Behind this landing capability is the full chain closed-loop of "data model ontology scene" constructed by cross dimensional intelligence. In June 2026, Cross Dimensional released the DexBEV embodied unified spatial representation framework, introducing the BEV bird's-eye view paradigm from the field of autonomous driving into the field of robotics; Aligned DexWorld, launched at the same time, has become the world's first large-scale spatiotemporal action alignment dataset that connects "human simulation real machine". The self-developed simulation engine DexVerse, which made its debut at WAIC 2026 ™， It is also the only embodied intelligent simulation engine in the world that supports multi physics field coupling in the same scene. This technology foundation enables cross dimensional to have the ability to continuously iterate and generate new energy in real industrial scenarios.

In Chen Hong's view, the current embodied intelligence industry is flourishing, but the true commercial scale has not yet been fully implemented. The cross dimensional approach is to 'lay eggs along the way' - using more than 50 real landing scenarios as a 'petri dish' for the evolution of general algorithms, and feeding back the brain's generalization ability with complex practical tasks, "Chen Hong said." The Data Expo is a good window, and we hope that in Guiyang, a computing hub city, more people can see that embodied intelligence is not just a distant concept - it has gradually begun to be commercialized from the laboratory. Looking back at the past two years, embodied intelligence stumbled, but now it has begun to appear in various industries with intensive applications. Models cannot do without computing power. Guiyang, as a computing node city in China, has a very good foundation and is very suitable for this industry.

Reporter Liu Yunsong