California-based Figure AI announced Wednesday that it is officially retiring its Figure 02 (F.02) humanoid robot.
The decision comes after an 11-month implementation at BMW Manufacturing’s Spartanburg, South Carolina, plant. This pilot project was part of a collaboration to test humanoid robots on a real assembly line.
The company highlighted that the F.02 unit contributed to the production of over 30,000 BMW X3 vehicles and the installation of over 90,000 sheet metal parts during its introduction.
Figure CEO Brett Adcock shared a visual of a robot covered in scratches, scrapes and dirt, hinting at the reality of working in an industrial environment.
Scars are proof of real work
The company shared video footage of its F.02 robot that clearly shows wear and tear from months on the line. Adcock described this as a “real-world expansion.”
This visual serves as proof against earlier skepticism that Mr. Figure’s work at BMW was nothing more than a small-scale feasibility study. By showing the robot’s well-worn appearance, the company demonstrates that the robot has been on an active assembly line for several months.
Dirt and scratches became unintended badges of honor. These show that the robots withstood the repetitive and demanding nature of factory work. The company said visual and performance metrics taken together validate the long-term adoption claim.
Performance data from trial run
The US company has released a detailed explanation of what the robots accomplished at the facility. After an initial setup period, adoption expanded rapidly, he said. Within the first six months, the robot was moved to Spartanburg and operational on site. By the 10th month, they were working full shifts on the assembly line.
Their main job was to lift the sheet metal parts out of the bin and place them in a welding fixture with a 5 millimeter tolerance. After placement, a conventional robotic arm performed the welding. The humanoid handled the metal load with a cycle time of 84 seconds, including 37 seconds of loading. Accuracy remained above 99 percent, the company said in a press release.
The robot completed over 1,250 hours of run time. The company estimated that the machine walked approximately 320 miles through the facility. The implementation followed a 10-hour shift schedule, Monday through Friday. These metrics were presented as evidence that humanoid robots can sustain industrial workloads for long periods of time in operating plants.
Lessons learned and the future of humanoids
The robotics company was also transparent about the challenges with the hardware it was implementing. The forearm emerged as a major point of failure due to the complexity of incorporating three degrees of freedom, thermal management, and cabling into a human-sized limb. Constant movement puts stress on microcontrollers and wiring, but this issue is rarely addressed in the humanoid field.
These lessons influenced the design of Figure 03. The new model eliminates the wrist switchboard and dynamic cabling, allowing the motor controller to communicate directly with the main computer.
The retirement of F.02 marks the transition from pilot testing to large-scale production. The company says, “Figure 02 taught us early lessons about what is needed for shipping.”
The fleet retirement paves the way for Figure 03, which the company claims is ready for large-scale implementation.
