테슬라 뉴스
Tesla's Next Leap: Optimus V3's Revolutionary Hand and Arm Design Revealed in New Patents
~에 의해
Rio
~에
Apr 16, 2026
Quick Summary: Tesla Optimus V3 Hand & Arm Patents
-
Filing date: Same day as Tesla's "We, Robot" event — October 2024; published internationally in 2026
-
Three patents: "Mechanically Actuated Robotic Hand" / "Robotic Appendage" / "Joint Assembly for Robotic Appendage"
-
Architecture: Cable-and-tendon driven; actuators relocated to forearm (not fingers/palm) — lighter, faster, more precise hand
-
Dexterity: 22 degrees of freedom total (4 DoF per finger + 2 DoF wrist yaw/pitch); 3 control cables per finger
-
Key innovation: Specialized wrist cable transition mechanism — minimizes tendon length/tension changes during combined yaw+pitch movements; reduces stretch, friction, and crosstalk
-
Manufacturing intent: Simplified, stackable joint parts; modular design explicitly engineered for high-volume production — not a lab prototype
-
Musk's context: Hand = "majority of the engineering difficulty of the entire robot"; ~60% of overall Optimus challenge; by early 2026, Musk confirmed the "hardest problems" had been overcome
-
Status: Optimus showcased at AWE 2026 in Shanghai; V3 unveiling teased by Tesla China
A series of newly published international patents have revealed the design of what is believed to be the Optimus V3 hand and arm — a cable-and-tendon driven system with 22 degrees of freedom, actuators relocated to the forearm, and a specialized wrist routing mechanism that solves one of robotics' most persistent engineering problems. Filed on the same day as Tesla's "We, Robot" event in October 2024, these patents represent the transition from ambitious prototype to production-intent system. Here's the full technical breakdown.
The Three Patents: What Each Covers
| Patent |
Scope |
Key Innovation |
| Mechanically Actuated Robotic Hand |
Primary patent — the complete hand architecture |
Cable-and-tendon system with actuators in forearm; 22 DoF total; 3 control cables per finger routed through integrated phalange channels |
| Robotic Appendage |
Entire forearm-to-finger assembly as a cohesive system |
Formalizes how palm couples to forearm; how precise cable tensioning enables wide range of grips and gestures across the full limb |
| Joint Assembly for Robotic Appendage |
Micro-level joint mechanics |
Curved contact surfaces + composite flexible member = smooth, low-friction pivoting with consistent tension; simplified stackable parts designed for high-volume manufacturing |
The Core Architecture: Why Actuators in the Forearm?
| Design Choice |
Engineering Rationale |
Biological Parallel |
| Actuators relocated to forearm |
Removes heavy motors from fingers and palm — significantly lighter hand with lower inertia; enables faster, more precise, more energy-efficient movements |
Human forearm muscles control finger movements via tendons — Tesla's design directly mirrors this biological architecture |
| 3 control cables per finger |
Thin, flexible cables routed from forearm actuators through wrist and palm into finger segments; integrated phalange channels guide cables behind some joints and in front of others for independent joint control |
Human finger tendons routed through tendon sheaths — same principle of selective tension application |
| 22 total degrees of freedom |
4 DoF per finger (complex bending/curling) + 2 DoF wrist (yaw + pitch) = 22 DoF total; approaches human hand dexterity |
Human hand: 27–28 DoF — Optimus V3 reaches approximately 79% of human hand articulation |
The Key Innovation: Wrist Cable Transition Mechanism
The problem it solves: In tendon-driven systems, the wrist is a notorious failure point. As the wrist bends and rotates, tendons stretch, rub against each other, and generate friction — causing imprecise, jerky motions and eventual mechanical failure. This is the primary reason most tendon-driven robotic hands have limited reliability and precision.
| Mechanism Element |
How It Works |
What It Eliminates |
| Lateral-to-vertical cable stack transition |
Cables organized in lateral stack on forearm side; transitioned to vertical stack on hand side through a meticulously designed transition zone within the wrist |
Minimizes changes in tendon length and tension during combined yaw + pitch wrist movements |
| Geometry-optimized routing |
Unique routing geometry engineered specifically for multi-axis wrist operation — accounts for the complex 3D path cables must travel during simultaneous yaw and pitch |
Dramatically reduces cable stretch, torque, friction, and crosstalk between tendons — the four primary failure modes of tendon-driven wrists |
| Result |
Smoother, more predictable, highly reliable multi-axis wrist operation |
Enables complex tasks from factory floor sorting to delicate home environment object handling — the full range of general-purpose humanoid use cases |
Musk's Years-Long Obsession: The Hand Problem in Context
| Timeline |
Musk's Statement / Status |
| Ongoing |
Hand = "the majority of the engineering difficulty of the entire robot"; ~60% of overall Optimus challenge; "harder than Cybertruck or Model X — somewhere between Model X and Starship" |
| Mid-2025 |
Musk candidly acknowledged Tesla was "struggling" to finalize the hand and forearm design — a rare admission of difficulty |
| Early 2026 |
Musk announced the company had overcome the "hardest" problems: human-level manual dexterity, real-world AI integration, and scalability for volume production |
| April 2026 |
Patents published; Optimus showcased at AWE 2026 in Shanghai; V3 unveiling teased by Tesla China
|
From Prototype to Production: The Manufacturing Intent
| Design Element |
Manufacturing Signal |
| Simplified, stackable joint parts |
Visible in patent diagrams; explicitly designed for ease of assembly — not a complex bespoke prototype but a modular system that can be assembled reliably at scale |
| No existing supply chain — invented from scratch |
Tesla has had to create the supply chain for these precision components; the patent filings protect the IP of components that will need to be manufactured at unprecedented volume |
| Giga Texas ramp context |
Musk noted that Cybercab and Optimus share the "almost everything is new" status at Giga Texas — both ramping simultaneously; the hand patents are the IP foundation for that production ramp |
| The competitive distinction |
Most humanoid competitors showcase impressive demos in controlled environments; Tesla is solving the manufacturing problem in parallel with the technology problem — the patents are evidence of this dual-track approach |
Conclusion
Key Takeaways
-
Three patents: Mechanically Actuated Robotic Hand + Robotic Appendage + Joint Assembly — filed October 2024, published 2026
-
Architecture: Tendon-driven; actuators in forearm; 22 DoF; 3 cables per finger — bio-inspired, lighter, faster, more precise than motor-in-finger designs
-
The key innovation: Wrist cable transition mechanism — lateral-to-vertical stack routing eliminates stretch, friction, and crosstalk during multi-axis wrist movement
-
Manufacturing intent: Simplified stackable joints; modular design; no lab prototype — engineered for high-volume production from first principles
-
Musk's milestone: Hand = ~60% of Optimus challenge; "struggling" in mid-2025; "hardest problems overcome" by early 2026
-
Current status: Optimus at AWE 2026 Shanghai; V3 unveiling imminent; Giga Texas production ramp underway
These patents are the strongest evidence yet that Tesla has not only understood the hardest problem in humanoid robotics but has engineered a production-ready solution. The 22-DoF tendon-driven hand with its crosstalk-minimizing wrist mechanism is not a research achievement — it is a manufacturing blueprint. Combined with the Giga Texas production ramp and the imminent V3 unveiling, Tesla is positioning Optimus to be the first humanoid robot that is both highly capable and manufacturable at the scale required to matter economically.
