Wheeled Robots Gain Ground as Embodied AI Moves from Labs to Work Floors

At recent industry shows, the machines actually moving boxes, sorting items and tending production lines are more often wheeled platforms with human-like upper bodies than bipedal humanoids. The shift reflects engineering priorities: stability, cost and measurable task performance now outweigh the pursuit of legged generality.

NextFin News — At the 2026 World Artificial Intelligence Conference, exhibition halls filled with robots performing practical work. Galbot’s S1 lifted and stacked 30-kilogram crates. Other systems sorted household items or tended miniature production lines. Across the floor, the majority of machines handling continuous material movement used wheels rather than legs.

Industry tallies of new humanoid-style products released in the first half of 2026 show wheeled platforms slightly outnumbering bipedal ones for the first time. One alliance tracking the sector counted more than 90 new full-machine releases, with wheeled designs accounting for roughly 48 percent and bipedal designs about 45 percent. Companies including Galbot, Pudu, Fourier Intelligence, Qianxun and Youibot introduced or highlighted wheeled dual-arm models aimed at industrial and commercial tasks. Youibot announced its “Xifeng” series with list prices of 299,000 yuan for the standard version and 499,000 yuan for the advanced version, along with stated intent orders for 4,000 units.

The preference is not aesthetic. In factories, warehouses, supermarkets and hotels, floors are largely flat. A mature wheeled chassis already provides reliable locomotion, high payload capacity, long operating times and relatively low maintenance. Designers can therefore concentrate computing power, sensors and algorithms on vision, dual-arm coordination and task execution rather than on continuous balance. Bipedal systems must maintain stability while moving, avoiding obstacles and manipulating objects; any shift in gait or center of gravity can degrade arm precision. Wheeled platforms separate those problems.

Several manufacturers describe the choice in practical terms. Fourier has positioned its new wheeled dual-arm GRW as a complement to its bipedal line, noting that wheeled products reach customers faster and suit heavy or repetitive handling at controllable cost. Qianxun’s leadership has pointed to shared chassis technology and supply chains across models as a way to cut development time and manufacturing expense while reusing perception and control software. Youibot, which reports hundreds of prior industrial deployments, emphasizes that new machines can inherit existing scheduling and software systems, reducing the data and time required to reach usable task success rates.

The limitations are equally clear. Wheels struggle with stairs, thresholds and uneven ground. Many real-world environments still contain such obstacles, and human-scale buildings often assume legged movement. Bipedal robots retain an advantage where those conditions dominate. Some developers therefore explore hybrid wheel-leg designs that switch modes according to terrain, seeking both efficient travel on flat surfaces and the ability to climb when needed.

Yet commercial value depends on whether customers will pay for extra mobility. Additional joints, more complex control and higher energy use raise cost and potential failure points. Only when crossing stairs or rough ground is an unavoidable part of the job does that extra capability justify the expense. For the large volume of indoor material handling, picking and tending tasks, the simpler wheeled form currently delivers clearer returns.

The industry is unlikely to converge on a single body type. Forklifts have never needed legs; fixed robot arms have never needed mobility. Each form persists where it solves a specific problem more efficiently than alternatives. Wheeled platforms with human-like upper bodies occupy the middle ground that is most accessible today: they can move between stations, grasp and place objects, and operate under existing factory or warehouse layouts without requiring a redesign of the environment.

Embodied intelligence will eventually demand greater generality. Until models, sensors and actuators reliably handle open-ended environments, the measure of progress remains whether a machine can enter a real workplace, complete defined tasks with quantifiable reliability, and do so at a cost customers accept. On that narrower but immediate test, wheels are proving more practical than legs.

本文系作者 Chelsea_Sun 授权钛媒体发表,并经钛媒体编辑,转载请注明出处、作者和本文链接
本内容来源于钛媒体钛度号,文章内容仅供参考、交流、学习,不构成投资建议。
想和千万钛媒体用户分享你的新奇观点和发现,点击这里投稿 。创业或融资寻求报道,点击这里

敬原创,有钛度,得赞赏

赞赏支持
发表评论
0 / 300

根据《网络安全法》实名制要求,请绑定手机号后发表评论

登录后输入评论内容

扫描下载App