BMW Tests Two Humanoid Robot Platforms: Figure 02 Moves 90,000 Parts, Hexagon AEON Rolls Into Leipzig

BMW is doubling down on humanoid robots. The German automaker announced last week that it will pilot Hexagon Robotics' wheeled humanoid, AEON, at its Leipzig plant in Germany, just months after wrapping a 10-month deployment of Figure AI's Figure 02 at its Spartanburg facility in South Carolina.

The numbers from Spartanburg are striking: Figure 02 moved more than 90,000 components, took approximately 1.2 million steps, and logged roughly 1,250 operating hours across 10-hour shifts, five days a week. During that time, the robot supported production of over 30,000 BMW X3 vehicles. BMW now says it is evaluating the next-generation Figure 03 for expanded use cases.

Illustration

A wheeled humanoid on the factory floor. BMW is testing multiple humanoid form factors across its global plants.

Two Humanoids, Two Philosophies

BMW's approach is unusual in the humanoid robotics space. Rather than betting on a single vendor, the company is running parallel evaluations with fundamentally different robot architectures. Figure 02 is a fully bipedal humanoid that walks on two legs and uses dexterous hands to manipulate parts. AEON, by contrast, is a semi-humanoid: it has two legs, but each leg ends in a wheel rather than a foot. The result is a robot that can roll across flat surfaces for speed and step over obstacles when needed.

The logic behind testing both makes sense for a company with dozens of plants worldwide. Assembly lines vary. Some environments demand the agility and human-like reach of a bipedal system. Others, particularly logistics corridors and battery assembly areas, benefit more from a fast-moving wheeled platform that can carry scanning tools and interchangeable grippers.

Under the Hood: Figure 02 vs. Hexagon AEON

The Figure 02 deployment at Spartanburg targeted the body shop, an area BMW chose deliberately because it already had a high degree of automation. Workers there had experience integrating new technologies, and material supply was already handled by automated smart transport robots (STRs). Figure 02's job was specific: removing and positioning sheet metal parts for the welding process. BMW described this task as demanding in both speed and accuracy, and physically exhausting for human workers.

According to BMW, Figure AI was able to quickly transfer motion sequences learned in lab environments to stable shift operations on the production floor. The robot was integrated into BMW's Smart Robotics ecosystem using standardized interfaces, which BMW credits for the faster-than-expected transition from testing to actual production work.

Illustration

Figure 02 handled sheet metal positioning for the welding process during its 10-month deployment at BMW Spartanburg.

Hexagon's AEON takes a different approach. Released in June 2025, the robot went through a theoretical evaluation phase and laboratory testing before BMW started its initial test deployment at Leipzig in December 2025. A second test deployment is planned for April 2026, with the full pilot phase beginning in summer 2026. BMW is focusing on AEON's modular design, which allows different hand elements, gripper types, and scanning tools to be swapped in depending on the task.

Who's Deploying Humanoids in Manufacturing

BMW is not the only automaker investing in humanoid robots, but it is arguably the most aggressive in testing multiple platforms simultaneously. The broader industry is moving quickly.

What BMW's Dual Strategy Means for the Industry

BMW's decision to test both bipedal and wheeled humanoids sends a specific signal to robot manufacturers: no single form factor will dominate factory floors. The company is building internal expertise across architectures, and its Smart Robotics ecosystem with standardized interfaces suggests it wants the ability to swap vendors and robot types without rearchitecting its production IT.

For robot makers, this is both an opportunity and a challenge. Companies like Figure AI, Hexagon, Agility, and Boston Dynamics are all competing for the same factory floor real estate. But BMW's willingness to run multiple pilots simultaneously means the market may support several winners, each optimized for different tasks. The body shop demands dexterity and bipedal mobility. Battery assembly needs speed and tool modularity. Logistics corridors favor wheeled platforms.

Milan Nedeljkovic, BMW's Board member for Production, framed it in terms of competitiveness: "The symbiosis of engineering expertise and artificial intelligence opens up entirely new possibilities in production." That language is measured, not hype. BMW is treating humanoid robotics as an engineering integration problem, not a moonshot. And the data from Figure 02 (90,000 components, 30,000 vehicles supported) gives them concrete evidence that the economics can work.

Meanwhile, the cautionary tale of K-Scale Labs (which shut down in late 2025 after failing to close its Series A) offers a counterpoint. As former COO Rui Xu wrote in a detailed post-mortem published today, the startup fell into what he calls "Large Model Chauvinism," the belief that AI models are getting so good that hardware can afford to be simple. K-Scale's collapse highlights the gap between demo-stage robots and production-ready ones. BMW's results with Figure 02 show what it takes to cross that gap: standardized interfaces, occupational safety integration, experienced factory teams, and thousands of hours of real-world operation.

Frequently Asked Questions

The 12-Month Outlook

BMW's humanoid strategy will hit several milestones in the next year. The Hexagon AEON pilot should be fully operational at Leipzig by mid-summer 2026. The evaluation of Figure 03 for expanded use cases at Spartanburg (and potentially other plants) will determine whether BMW scales its bipedal deployment from one task to many. If Hyundai follows through on its plans to purchase thousands of Boston Dynamics Atlas units, and Toyota expands its Agility Digit deployment in Canada, the second half of 2026 could mark the point where humanoid robots shift from pilot projects to permanent fixtures on automotive production lines.

The real question is not whether humanoid robots can work in factories. BMW just proved they can. The question is how fast automakers can identify the right robot for the right task, integrate it with existing IT systems, and scale deployment across a global plant network.