Figure AI runs 8‑hour shifts
What happened
- Figure AI showcased its Helix‑02 humanoid robots completing continuous package‑sorting shifts for more than eight hours while matching human speed in demonstrations. - The demo included multi‑robot coordination and a live‑streamed run that Figure says operates autonomously in warehouse tasks. - If sustained in production, the capability targets logistics automation and may pressure hiring and deployment timelines in fulfillment centers. (x.com/IntEngineering/status/2054947736387022954)
Why it matters
1/ Figure AI just demoed its Helix-02 humanoid robots sorting packages for over 8 hours straight—matching human speeds in a live-streamed warehouse task. No human intervention. Multi-bot coordination included. 2/ The May 14 livestream showed 6 Helix-02 units tackling inbound tote sorting: grabbing mixed packages, scanning barcodes, placing into outbound bins. Robots hit 99.9% accuracy, per Figure's post-demo metrics. Cycle time? 4.5 seconds per item, human parity. 3/ Key tech: Helix-02's upgraded arms with 16 DoF hands, force/torque sensing, and vision-language models for real-time decisions. No teleop—full autonomy via end-to-end RL trained on 100k+ hours of human demo data. Figure claims 10x sim-to-real transfer efficiency. 4/ Multi-robot sync relied on a central fleet manager handling collision avoidance and task allocation. During the 8+ hour run (8h 17m total uptime), bots swapped batteries autonomously without halting ops. Zero errors on 1,200+ packages. 5/ Why warehouses? Figure targets $100B logistics automation market. Early partner BMW tested Helix in SC factories last year; now expanding to fulfillment. Demo proves readiness for 24/7 shifts vs. human 8-hour limits. 6/ Speed breakdown: Inbound scan (1.2s), grasp (1.1s), place (1.2s), outbound confirm (0.9s). Matched Figure's internal human benchmark from 50 operators. Energy use? 2.5 kWh per robot-hour, 40% below prior gen. 7/ Backers: OpenAI ($675M round lead), Microsoft, Nvidia. Valuation hit $2.6B post-Series B. Figure's raised $675M total, deploying 100+ bots in production pilots by Q4 2026. 8/ Competitors: Tesla Optimus (Gen2 walks but no 8h demos yet), Agility Robotics Digit (Amazon pilot, tote-handling focus), Apptronik Apollo (NASA-funded). Figure differentiates on full autonomy + generalist skills. 9/ Roadmap: Helix-03 Q1 2027 with 20 DoF dexterous hands, 2x speed. 1,000-bot fleet for Amazon/ UPS-style clients by 2028. Cost target: $20k/unit at scale (down from $100k prototypes). 10/ Watch the full stream: Robots never stopped, even during a mock "lunch break" where humans would've paused. This isn't hype—it's the first public proof humanoid labor scales to shift-length endurance.
Key numbers
- Figure AI showcased its Helix‑02 humanoid robots completing continuous package‑sorting shifts for more than eight hours while matching human speed in demonstrations.
- (x.com/IntEngineering/status/2054947736387022954) 1/ Figure AI just demoed its Helix-02 humanoid robots sorting packages for over 8 hours straight—matching human speeds in a live-streamed warehouse task.
- 2/ The May 14 livestream showed 6 Helix-02 units tackling inbound tote sorting: grabbing mixed packages, scanning barcodes, placing into outbound bins.
- Robots hit 99.9% accuracy, per Figure's post-demo metrics.
What happens next
- 2/ The May 14 livestream showed 6 Helix-02 units tackling inbound tote sorting: grabbing mixed packages, scanning barcodes, placing into outbound bins.
- Figure targets $100B logistics automation market.
- Cost target: $20k/unit at scale (down from $100k prototypes).
Quick answers
What happened in Figure AI runs 8‑hour shifts?
Figure AI showcased its Helix‑02 humanoid robots completing continuous package‑sorting shifts for more than eight hours while matching human speed in demonstrations. The demo included multi‑robot coordination and a live‑streamed run that Figure says operates autonomously in warehouse tasks. If sustained in production, the capability targets logistics automation and may pressure hiring and deployment timelines in fulfillment centers. (x.com/IntEngineering/status/2054947736387022954)
Why does Figure AI runs 8‑hour shifts matter?
1/ Figure AI just demoed its Helix-02 humanoid robots sorting packages for over 8 hours straight—matching human speeds in a live-streamed warehouse task. No human intervention. Multi-bot coordination included. 2/ The May 14 livestream showed 6 Helix-02 units tackling inbound tote sorting: grabbing mixed packages, scanning barcodes, placing into outbound bins. Robots hit 99.9% accuracy, per Figure's post-demo metrics. Cycle time? 4.5 seconds per item, human parity. 3/ Key tech: Helix-02's upgraded arms with 16 DoF hands, force/torque sensing, and vision-language models for real-time decisions. No teleop—full autonomy via end-to-end RL trained on 100k+ hours of human demo data. Figure claims 10x sim-to-real transfer efficiency. 4/ Multi-robot sync relied on a central fleet manager handling collision avoidance and task allocation. During the 8+ hour run (8h 17m total uptime), bots swapped batteries autonomously without halting ops. Zero errors on 1,200+ packages. 5/ Why warehouses? Figure targets $100B logistics automation market. Early partner BMW tested Helix in SC factories last year; now expanding to fulfillment. Demo proves readiness for 24/7 shifts vs. human 8-hour limits. 6/ Speed breakdown: Inbound scan (1.2s), grasp (1.1s), place (1.2s), outbound confirm (0.9s). Matched Figure's internal human benchmark from 50 operators. Energy use? 2.5 kWh per robot-hour, 40% below prior gen. 7/ Backers: OpenAI ($675M round lead), Microsoft, Nvidia. Valuation hit $2.6B post-Series B. Figure's raised $675M total, deploying 100+ bots in production pilots by Q4 2026. 8/ Competitors: Tesla Optimus (Gen2 walks but no 8h demos yet), Agility Robotics Digit (Amazon pilot, tote-handling focus), Apptronik Apollo (NASA-funded). Figure differentiates on full autonomy + generalist skills. 9/ Roadmap: Helix-03 Q1 2027 with 20 DoF dexterous hands, 2x speed. 1,000-bot fleet for Amazon/ UPS-style clients by 2028. Cost target: $20k/unit at scale (down from $100k prototypes). 10/ Watch the full stream: Robots never stopped, even during a mock "lunch break" where humans would've paused. This isn't hype—it's the first public proof humanoid labor scales to shift-length endurance.
