BMW is preparing to expand its work with humanoid robots at Plant Spartanburg in South Carolina. The company has announced a new project involving Figure AI’s Figure 03 robot, shifting the focus from a previous body-shop application toward logistics sequencing. The planned task is specific: picking unsorted components from larger containers and arranging them in the correct order inside sequencing trolleys for use in production.
The project is another test of what the automotive industry increasingly calls Physical AI—the combination of artificial intelligence, sensors and machines that can perceive and act in real industrial environments. BMW presents the initiative as part of its broader iFACTORY strategy, but the announcement should be read as a defined project rather than proof that humanoid robots are ready for unrestricted factory deployment.
From body-shop handling to logistics sequencing
BMW’s earlier work with Figure 02 concentrated on a repetitive production task. According to the company, the robot inserted sheet-metal parts into fixtures for welding during the production of more than 30,000 BMW X3 vehicles over a ten-month period. That experience gave BMW and Figure a practical environment in which to study robot movement, integration with manufacturing processes and interaction with existing equipment.
The Figure 03 project moves into logistics. Sequencing is essential in modern vehicle manufacturing because the right components must reach the line in the right order and at the right time. A robot working in this area must identify parts inside a mixed container, grasp them reliably and place them correctly in a trolley. The challenge is not a single dramatic movement but consistent performance across many slightly different situations.
What Figure 03 is expected to do
BMW says Figure 03 will pick unsorted components from larger containers and sort them into sequencing trolleys. This requires perception, object recognition, grasp planning and controlled placement. Components may overlap, change orientation or be partly hidden, so the machine must adjust rather than repeat one fixed motion.
The wording matters. BMW has announced the project and described the intended task; it has not claimed that Figure 03 has already completed a full production deployment at Spartanburg. The next phase will show how the robot performs under factory conditions, including cycle-time demands, changing component positions and the need to operate safely around people and other equipment.
Lessons from the Figure 02 deployment
The earlier Figure 02 deployment provides the immediate background. BMW says the robot supported production of more than 30,000 X3 vehicles over ten months by inserting sheet-metal parts for welding. This is a company-reported result from a particular application, not a general validation of humanoid robots across the automotive sector.
Even so, the continuity between the two projects is significant. Instead of treating the humanoid as a one-off demonstration, BMW is applying lessons from one controlled task to another. The move also highlights why manufacturers are interested in a human-shaped machine: it may be able to operate in spaces, reach workstations and handle equipment designed around human workers without rebuilding an entire production area.
Hardware changes aimed at factory work
BMW lists several Figure 03 features that are relevant to industrial use. These include soft components, tactile sensors, cameras in the palms, wireless charging and audio communication. Tactile sensing and palm cameras can help the robot understand contact and see objects during close manipulation. Softer external parts may reduce some risks during incidental contact, although safe deployment still depends on the full system and operating procedure.
Wireless charging could allow the robot to replenish energy without a worker manually connecting a cable, while audio communication may support more natural interaction with nearby staff. These features indicate a design aimed at longer, more integrated operation. They do not by themselves establish reliability, safety or economic value; those questions depend on measured performance in the actual process.
Where humanoids fit beside existing automation
Automotive plants already use highly capable industrial robots, conveyors, automated guided vehicles and specialized handling systems. Humanoids are unlikely to replace that infrastructure wholesale. Their potential advantage is flexibility in areas where products and tasks vary, workspaces were designed for people, or conventional automation would require expensive custom engineering.
Logistics sequencing is therefore a useful test. It combines physical handling with variation and decision-making, while remaining narrow enough to measure. BMW can compare accuracy, speed, uptime, intervention requirements and safety performance against existing methods. Those operational details will ultimately matter more than the robot’s appearance.
The project also illustrates a broader shift in robotics. Better vision models, simulation tools and learning systems are being connected to machines that must deal with friction, imperfect objects and unpredictable surroundings. Progress in software is important, but factory value depends on the complete system: hardware durability, maintenance, integration, worker training and a clear reason to automate the task.
RoviVox View
BMW’s Figure 03 project is notable because it follows an earlier production application and targets a concrete logistics problem. That makes it more informative than a stage demonstration. However, it remains one company’s planned project in one facility, and its results should not be generalized before BMW publishes operational evidence.
The most useful question is not whether humanoids can enter factories, because they already have in limited trials. It is whether they can deliver repeatable value beside mature automation without adding unacceptable cost or complexity. The Spartanburg sequencing project could provide a clearer answer if BMW reports measurable outcomes after deployment.