Figure AI’s latest humanoid robot has reportedly walked 348 miles without leaving a treadmill.
On Oct. 19, Figure CEO Brett Adcock reported that the company’s Figure 03 humanoid had logged approximately 1.2 million steps over nine days during an endurance experiment. According to Adcock, the test began as a side project intended to stress the robot’s hardware and expose potential mechanical failures.
The milestone offers another glimpse into the progress of humanoid robotics. But for manufacturers and warehouse operators, the bigger question is whether impressive results in controlled environments can translate into reliable performance during everyday operations.
Figure 03 completes 1.2 million steps in treadmill endurance test
Adcock shared the results on X on Oct. 19, along with footage of Figure 03 walking on a treadmill inside a testing facility, according to Interesting Engineering.
The company reported three notable figures:
- 348 miles (560 kilometers): The total distance Figure said its robot covered during the treadmill experiment.
- 1.2 million steps: The approximate number of steps reported by Adcock.
- Nine days: The reported duration of the walking test.
The footage shows Figure 03 walking with a cable attached. However, Figure has not disclosed the cable’s exact purpose or how the robot’s power requirements were managed throughout the experiment.
The company has also not publicly clarified whether a single robot completed the entire test without replacement, whether maintenance was required, or whether the experiment involved interruptions. Although Adcock described the robot as having walked for nine consecutive days, the available information does not independently establish that it operated uninterrupted.
These distinctions matter. The results should not be interpreted as evidence that Figure 03 can operate for nine days on a single battery charge or independently navigate complex environments for that duration.
Still, the reported distance provides a useful indication of the repetitive mechanical demands humanoid robots must withstand before widespread commercial deployment becomes practical.
What the 348-mile test reveals about humanoid robot reliability
Walking hundreds of miles on a treadmill may seem less impressive than watching a robot climb stairs or manipulate objects. However, repetitive movement creates its own engineering challenges.
Humanoid robots depend on motors, actuators, joints, sensors, and control systems working together to maintain balance and produce coordinated movement. Extended walking tests can help engineers identify problems related to component wear, mechanical stress, thermal management, and movement consistency.
Those considerations become particularly important when robots are expected to operate regularly in demanding industrial environments.
Figure has previously demonstrated more complex capabilities. In September, Figure 03 attempted household tasks across 30 previously unseen homes, successfully completing 237 of 420 trials, according to the company. That experiment demonstrated progress in adapting robotic skills to unfamiliar environments, but it also highlighted the reliability challenges that remain before humanoid robots can consistently perform everyday tasks.
The company has also conducted extended warehouse demonstrations.
In May, TechRepublic reported on Figure’s humanoid robots sorting packages during a livestream that had reached approximately 40 hours and 50,000 packages.
The demonstration subsequently continued. Figure later reported that the rotating fleet had operated for approximately 200 hours and collectively processed 249,560 packages. Ars Technica also documented the extended experiment.
According to the reported results in the figure, the machines rotated through the task, allowing individual robots to recharge between shifts. Each reportedly operated for approximately three to four hours before another took over.
That makes the two experiments fundamentally different.
The warehouse demonstration focused on completing repetitive work through a rotating fleet, while the latest treadmill experiment emphasizes accumulated walking distance and mechanical endurance.
Neither result alone establishes the long-term reliability required for routine commercial deployment.
More must-read AI coverage
- SS&C Intralinks DealCentre AI vs. Datasite: Which platform is built for the future of dealmaking?
- SS&C Intralinks FundCentre AI vs. Juniper Square: Which platform better supports modern private markets fund managers?
- Why Data, Not Models, Determines AI Success
- The Rise of the AI-Native Factory: How Physical AI Is Transforming Manufacturing
What the endurance test means for businesses considering humanoid robots
For warehouse operators, manufacturers, and IT teams evaluating humanoid robots, the 348-mile experiment highlights one part of a much larger reliability challenge.
A robot capable of repeatedly walking for extended periods may be better positioned to handle physically demanding environments. However, commercial value depends on more than movement.
Businesses also need evidence that robots can perform useful tasks consistently, recover from errors, operate safely around employees, and deliver measurable returns on investment.
Before committing to humanoid deployments, organizations should consider several questions:
- Operational reliability: Can the robot maintain consistent performance across repeated shifts without frequent repairs or human intervention?
- Real-world adaptability: Can it navigate obstacles, changing layouts, and unpredictable workplace conditions?
- Total operating costs: How do maintenance, electricity, supervision, and downtime affect deployment economics?
These questions are especially relevant as robotics companies move beyond controlled demonstrations toward commercial applications.
Figure is not alone in pursuing industrial applications for humanoid robots. In September, TechRepublic covered Agility Robotics’ Digit 5 humanoid, which the company is developing for repetitive material-handling tasks in warehouses and manufacturing facilities.
Figure’s treadmill experiment contributes another piece of evidence to the broader effort to make humanoid robots commercially practical.
However, businesses should distinguish between a robot’s ability to withstand repetitive movement and its ability to perform productive work reliably.
For IT leaders and operations executives, the most useful next steps will be to watch for independently verified endurance results, maintenance data, real-world uptime figures, and evidence that humanoid robots can complete extended shifts without frequent intervention.
Until then, 348 miles is a notable engineering milestone. The more consequential test will come when robots demonstrate comparable reliability while performing useful work outside controlled environments.
More robot news: For a closer look at the challenges of deploying humanoid robots in the workplace, read TechRepublic’s Humanoid Robots Are Getting Cheaper, But Enterprise Costs Are Just Getting Started, which examines battery life, infrastructure requirements, and operating costs.
Matt Gonzales
Matt Gonzales is a technology journalist, editor, and content strategist with more than a decade of experience covering emerging technologies, enterprise IT, cybersecurity, artificial intelligence, and workplace innovation. As Managing Editor for eWeek and TechRepublic, he leads editorial strategy and newsroom operations while helping business and IT leaders navigate an evolving technology landscape. Throughout his career, Matt has held leadership roles overseeing content development, editorial planning, and newsroom operations across digital publications and enterprise media organizations. Before joining TechnologyAdvice, he served as an editor at SHRM, where he covered workplace trends and emerging technologies, and as Lead Writer and Editor for Marine Corps Systems Command, where he reported on defense technologies, innovation initiatives, and government technology programs. Matt's expertise spans cybersecurity, enterprise technology, AI, B2B software, technical writing, and digital publishing. He has reported on major technology developments, including the rapid evolution of generative AI, helping readers understand both the opportunities and risks associated with emerging technologies. His work combines deep research, editorial rigor, and practical business insights to make complex technical topics accessible to a broad audience. An award-winning journalist, Matt has earned recognition for excellence in reporting and editorial leadership. He holds a Bachelor of Science in Communication with a concentration in Journalism from East Carolina University and continues to focus on delivering trusted analysis and actionable insights for technology, cybersecurity, and business professionals.