Skip to main content

Figure AI Humanoids: Neural Nets, Manufacturing, and a Trillion-Dollar Market

[HPP] Brett AdcockFebruary 12, 20266 min
29 connections·40 entities in this video→

The Evolution of Figure AI Robots

  • πŸš€ Figure's robots evolved from Figure 1 (proof of concept) to Figure 3 (production intent).
  • πŸ’‘ Figure 3 achieved significant improvements, including being 30 pounds lighter and reducing manufacturing costs by 90%.
  • ✨ New features on Figure 3 include palm cameras and tactile fingers for enhanced interaction.

Shift to Neural Network Control

  • 🧠 The software stack transitioned from hundreds of thousands of lines of handwritten C++ to an all neural net approach.
  • πŸ€– Helix is the core neural architecture, providing full-body reinforcement-learned control integrated with cameras and tactile sensors.
  • 🎯 This advanced control enables capabilities like bimanual manipulation, dynamic locomotion, and continuous replanning in complex, real-world environments.

Data Advantage and Generalization

  • πŸ“Š A key data advantage is that once one robot learns a task, every robot in the fleet knows it, creating a powerful network effect.
  • ⚠️ Handwritten heuristics proved unable to generalize to the combinatorial complexity of a humanoid body, making learning from diverse physical experience essential.
  • πŸ› οΈ Co-designed hardware and software (Figure 3 built for Helix) ensures onboard inference, allowing robots to operate without constant, reliable network connections.

Manufacturing, Business, and Impact

  • 🏭 Figure is establishing Baku production lines with the ambitious goal of achieving robots building robots and scaling commercial fleets this year.
  • πŸ’° The preferred leasing business model aims to stabilize fleet operations, align incentives, and democratize access to humanoid technology.
  • ❀️ Humanoids have profound social impact potential in areas like healthcare and elder care, reducing isolation and enabling aging in place.

Scaling, Safety, and Future Vision

  • πŸ“ˆ Training at scale involves heavy investment in clusters and pre-training data, with a vision for a single omni-model fusing language, vision, and physics.
  • βœ… Safety is a top priority, addressed through multiple protection layers, including physical safety, redundant hardware, cybersecurity, and transparent data governance.
  • 🌐 The market for general-purpose humanoids is projected to be a multi-trillion dollar market, fundamentally transforming global GDP and labor dynamics.
Knowledge graph40 entities Β· 29 connections

How they connect

An interactive map of every person, idea, and reference from this conversation. Hover to trace connections, click to explore.

Hover Β· drag to explore
40 entities
Chapters3 moments

Key Moments

Transcript25 segments

Full Transcript

Topics15 themes

What’s Discussed

Humanoid robotsNeural networksHelix architectureReinforcement learningData advantageOnboard inferenceManufacturing processesFleet operationsLeasing modelHealthcare applicationsElder careCybersecurityOmni-modelVertical integrationMulti-trillion dollar market
Smart Objects40 Β· 29 links
PeopleΒ· 2
CompaniesΒ· 3
ProductsΒ· 9
ConceptsΒ· 20
EventsΒ· 5
LocationΒ· 1