Historically, industrial automation solved manipulation through ruthless task specialization: rigid two-jaw pneumatic parallel grippers or high-flow vacuum suction cups mounted to fixed six-axis arms. However, general-purpose humanoids operating in environments designed for human hands cannot rely on task-specific clamps.

This necessity has created a deep mechanical and algorithmic divide in robotic hand architecture: 6-DoF underactuated hands versus 20+ DoF fully actuated hands.

Underactuated hands (exemplified by commercial platforms like the Inspire-Robots RH56 series or early industrial humanoid grippers) use mechanical linkages, differential pulleys, and passive springs to wrap multiple phalanges around irregular objects using fewer actuators than degrees of freedom.

At the opposite end of the spectrum, fully actuated hands (championed by 1X NEO’s 22-DoF tendon digits, Shadow Robot’s Dexterous Hand, and Tesla Optimus Gen 3’s forearm-actuated architecture) assign independent motors or dedicated active tendons to nearly every joint, striving for true biological human equivalence.

This engineering comparison examines the mechanics, mass distribution, control software overhead, tactile sensor integration, and manufacturing economics that govern both philosophies.

Key Architectural Takeaways

• Mechanical Definition Split: Underactuated hands feature fewer active motors than geometric degrees of freedom ($N_{actuators} < N_{DoF}$), relying on passive compliance to contour around objects; fully actuated designs maintain independent motor control over each joint axis ($N_{actuators} \approx N_{DoF}$).

• In-Hand Manipulation Bottleneck: 6-DoF underactuated hands excel at envelope power grasping but cannot perform in-hand reorientation (e.g., twirling a pen or adjusting a key); 20+ DoF hands provide full Cartesian fingertip trajectory control.

• Mass and Inertia Constraints: Housing 20+ micro-motors inside the palm adds severe distal mass and thermal buildup; modern 20+ DoF architectures must relocate prime movers into the forearm via tendon conduits.

• Control Complexity Differential: Underactuated hands simplify grasping into binary open/close motor commands with automatic mechanical adaptation; 20+ DoF hands demand high-dimensional reinforcement learning (RL) policies and real-time contact dynamics models.

• Commercial Reliability vs. Dexterity: Underactuated hands deliver high mean time between failures (MTBF) and cost under $3,000; 20+ DoF hands cost $15,000 to $80,000 and introduce fragile tendon calibration and micro-gear wear.

Quick Specs: Underactuated (6-DoF) vs. Fully Actuated (20+ DoF) Hands

Mechanics of Underactuation: Compliance via Physical Intelligence

Underactuated hands operate on a principle known as mechanical intelligence. Rather than requiring microprocessors to calculate inverse kinematics for 15 individual finger knuckles, the physical linkage automatically adapts to the contour of an unfamiliar object.

Underactuated Motion Pipeline: The Differential Linkage Principle

• A single DC motor drives an input crank or tendon cable connected to a finger module.

• The proximal, intermediate, and distal phalanges are linked via planar four-bar linkages, floating whiffletree bars, or spring-loaded mechanical stops.

• When the proximal link makes physical contact with an object, its forward rotation halts.

• Because the drive linkage is differential, the motor continues pulling; the remaining kinetic energy transfers forward to curl the distal and intermediate phalanges inward.

↓ (Mechanical Adaptation Loop)

Autonomous Passive Grasp Formation

• The finger naturally wraps itself around the geometry of the target—whether it is a soda can, an apple, or an irregular screwdriver handle.

• Contact forces are passively distributed across multiple inner pad surfaces without closed-loop sensor feedback.

• Spring returns automatically straighten the finger when the motor reverses.

In a typical 6-DoF underactuated hand, each of the four fingers utilizes one motor for flexion/extension, while the opposable thumb utilizes two motors (one for flexion, one for circumduction/rotation).

This setup provides sufficient dexterity to form all standard human prehensile grasp classes: cylindrical power grasp, spherical power grasp, hook grasp, lateral key pinch, and precision fingertip pinch. Crucially, the entire actuation module fits inside the hand chassis itself, eliminating the need to run dozens of delicate cables through the wrist joint.

The 20+ DoF Imperative: Why Industrial Scaling Demands Full Actuation

While underactuated hands provide reliable power grasps, they suffer from an insurmountable kinematic boundary: they cannot perform in-hand dexterous manipulation.

When a human worker picks up an electric drill, writes with a pencil, or threads a hex nut onto a bolt, the fingers do not remain locked in their initial contact positions. The thumb, index, and middle fingers perform finger gaiting—releasing contact, repositioning along the object’s surface, sliding, and tilting the object within the palm envelope.

Kinematic Limitation 1: The Underactuated Grasp Trap

• Once an underactuated finger contacts an object, its intermediate joint positions are dictated by external contact points.

• The robot cannot choose to lift the middle knuckle while keeping the fingertip pinned down.

• If an object slips or is acquired at an awkward angle, the robot must set the item down on a flat table, reposition its arm, and pick it up again.

↓ (Kinematic Freedom Expansion)

Kinematic Mastery 2: Fully Actuated 20+ DoF Dexterity

• Every individual knuckle (Metacarpophalangeal [MCP], Proximal Interphalangeal [PIP], Distal Interphalangeal [DIP]) possesses independent rotational control.

• Includes lateral digit splaying (adduction/abduction) across all five fingers, plus multi-axis opposition across the carpometacarpal (CMC) thumb saddle.

• Allows the hand to execute true in-hand reorientation: unscrewing caps, rotating keys in locks, re-orienting small electronic connectors, and picking single thin washers off flat metal plates.

For high-mix assembly tasks—such as inserting delicate automotive wire harness clips or assembling complex hydraulic fittings—20+ degrees of freedom are not an aesthetic luxury; they are the baseline mechanical requirement to replace human manual labor.

The Packaging Dilemma: Palm Servos vs. Forearm Tendon Conduits

Designing a 20+ DoF robotic hand forces a severe structural compromise: where do you place 20+ independent actuators?

Packaging Strategy 1: Self-Contained Palm Actuation (Modular but Constrained)

• Layout: Micro-brushless motors, planetary micro-gearboxes, and drive inverters are packed into the palm enclosure and proximal finger knuckles.

• Advantage: Modularity. The entire hand unbolts at the wrist via a standard flange, with power and CAN-FD/EtherCAT data routed through a single sealed connector.

• Severe Liabilities:

  • Thermal Trap: Squeezing 15 to 20 motors into a confined space causes heat build-up, altering internal sensor calibrations.

  • Distal Mass Burden: The hand becomes heavy (exceeding 1.2 kg), increasing rotational inertia at the end of the arm.

  • Kinematic Limits: Constrained palm volume forces designers to cut degrees of freedom, typically capping hand complexity around 16 DoF (as seen in Figure 02).

↓ (Spatial Mass Relocation)

Packaging Strategy 2: Forearm-Mounted Tendon Actuation (Biomimetic but Complex)

• Layout: The hand contains zero electric motors. All 20 to 25 prime movers are bundled in the forearm chassis (mimicking the human flexor digitorum muscles).

• Transmission: Forces transmit across the multi-axis carpal wrist joint via synthetic high-molecular-weight polyethylene (UHMWPE) tendons running through low-friction Bowden conduits.

• Advantage: The physical hand remains ultra-lightweight (under 500 grams), enabling rapid finger accelerations and human-scale slender digits.

• Severe Liabilities:

  • Tendon stretch and hysteresis require continuous tension monitoring and software recalibration.

  • Bowden conduits experience internal friction wear over millions of cycles.

  • Servicing a single snapped cable requires bench labor to re-thread the forearm-to-fingertip path.

Tactile Perception: From Binary Touch to 3D Slip Vectors

A fully actuated hand with 20+ degrees of freedom is useless without continuous tactile perception. If an autonomous controller cannot feel when a fingertip begins to lose friction, high motor power will either crush fragile components or drop heavy tools.

Tactile Processing Pipeline across Multi-DoF Hands:

1. Surface Pressure Acquisition (1,000 Hz)

• Flexible piezoresistive arrays, capacitive skins, or internal micro-cameras tracking elastomeric dome deflections (e.g., GelSight-class sensors) measure normal force vectors.

• Underactuated hands typically monitor simple total clamping force through motor current draw.

• 20+ DoF hands require dense spatial arrays (60 to 100+ taxels per fingertip) to determine exact local contact points.

↓ (Real-Time Vector Extraction)

2. Dynamic Micro-Slip Detection (Sub-2 Millisecond Latency)

• High-frequency vibrational micro-sensors detect the initial acoustic signature of an object sliding across the silicone skin pad.

• The controller calculates normal force ($F_n$) vs. tangential shear force ($F_t$) to estimate localized Coulomb friction coefficients.

↓ (Distributed Closed-Loop Actuator Response)

3. Coordinated Finger Trajectory Compensation

• In an underactuated hand, the only available reaction is increasing overall motor clamp torque, which risks crushing thin-walled packaging.

• In a 20+ DoF hand, the whole-body hand controller dynamically redistributes force: adjacent fingers adjust their contact angles, the thumb increases opposition pressure, and the wrist re-aligns to shift the object’s gravitational center of mass.

Operational Video Reference: Dexterous Hand Mechanics in Action

The mechanical difference between coupled, underactuated grasping and independent, fully articulated dexterity can be seen in detailed hardware testing:

Dexterous Hand Kinematics and Tactile Grasping Showcase:

Watch advanced robotic manipulators in action: Shadow Dexterous Hand with Tactile Sensing Demonstration

• Key Observation Points:

  • Complex finger gaiting and in-hand object manipulation using 20+ independent tendon lines.

  • Multi-axis thumb opposition wrapping around asymmetrical household and workshop tools.

  • Rapid tactile slip compensation preventing smooth objects from dropping during dynamic arm motion.

Commercial ROI: The $3,000 Workhorse vs. The $40,000 Specialist

The selection between underactuated and fully actuated hands is ultimately driven by the economics of industrial maintenance and task requirements.

Commercial Profile 1: 6-DoF Underactuated Hand (The Volume Production Winner)

• Procurement Cost: $2,000 to $5,000 in commercial volume.

• Component Simplicity: Uses 5 or 6 standard off-the-shelf micro-actuators and basic planar linkage pins.

• Maintenance Profile: Rugged. Impact forces are absorbed by passive mechanical springs. A line technician can swap an entire finger sub-assembly in under 15 minutes using standard metric tools.

• Ideal Workflows: Logistics tote loading, structured machine tending, bin picking of rigid parts, and standardized palletizing.

↓ (Task Complexity vs. Cost Expansion)

Commercial Profile 2: 20+ DoF Fully Actuated Hand (The Specialized Frontier)

• Procurement Cost: $25,000 to $60,000+ per pair.

• Component Complexity: Requires 20+ high-precision miniature motors, custom planetary or cycloidal micro-reducers, miniature tension load cells, and micro-machined internal wire routes.

• Maintenance Profile: High service overhead. Tendon lines stretch over time, requiring periodic re-tensioning, while miniature gears are vulnerable to stripping under lateral shock loads.

• Ideal Workflows: Complex mechanical assembly, flexible wire routing, delicate laboratory testing, unstructured domestic chores, and eldercare assistance.

Engineering Verdict & Field Evaluation

6-DoF Underactuated Hands: Pros & Operational Strengths

• Inherent Mechanical Compliance: Naturally adapts to varying object geometries without requiring complex vision or force-control algorithms.

• Exceptional Impact Durability: Linkage-and-spring mechanisms absorb sudden collisions without stripping delicate gear teeth.

• Accessible Unit Economics: Lower bill-of-materials cost makes fleet-scale deployment financially practical today.

6-DoF Underactuated Hands: Limitations & Engineering Risks

• Zero In-Hand Manipulation: Cannot reorient, twist, or adjust items within the grasp envelope; mistakes require setting the part down completely.

• Limited Pinch Precision: Tendency to force objects into full power grasps makes fine-tip extraction of small fasteners difficult.

20+ DoF Fully Actuated Hands: Pros & Operational Strengths

• True Human Dexterity: Independent joint control enables human-equivalent finger gaiting, tool twisting, and fine assembly.

• Dynamic Tactile Control: Supports delicate, distributed force regulation across individual contact pads, handling soft, brittle, or flexible parts safely.

• Long-Term Physical AI Foundation: The only architecture capable of mastering the full distribution of unstructured human manipulation tasks.

20+ DoF Fully Actuated Hands: Limitations & Engineering Risks

• Severe Mechanical Fragility: Complex tendon runs and miniature micro-gears create numerous points of failure under industrial shift wear.

• Prohibitive Hardware Cost: Complex manufacturing keeps fully actuated hands out of reach for cost-sensitive logistics deployments.

The Bot.to Benchmark Verdict:

For warehouse fulfillment, parcel sorting, and structured machine tending today, 6-DoF underactuated hands are the correct commercial engineering choice. They survive dropped loads, require negligible software overhead, and offer immediate return on investment without delicate maintenance cycles.

However, for the broader promise of humanoid robotics—replacing human labor in complex automotive assembly, flexible wiring installation, and unstructured domestic assistance—20+ DoF fully actuated hands are the mandatory future. Companies that master the mass production of reliable forearm-driven tendon systems and self-calibrating tactile skins will set the standard for high-dexterity robotic manipulation.

Frequently Asked Questions (FAQ)

Q: What is the main difference between an underactuated and a fully actuated robot hand?

A: An underactuated hand has fewer motors than joints (e.g., 5 or 6 motors driving 15 knuckles), using passive springs or linkages to wrap around objects automatically. A fully actuated hand has independent motors for nearly every joint (20+ motors for 20+ degrees of freedom), giving complete control over each finger’s exact trajectory.

Q: Why don’t all humanoid robots use 20+ DoF hands?

A: 20+ DoF hands are expensive ($25,000–$60,000+), mechanically fragile, heavy, and difficult to control. For basic factory tasks like moving plastic totes or picking rigid boxes, the extra dexterity is unnecessary, and simpler underactuated grippers offer higher durability at a fraction of the cost.

Q: What is in-hand manipulation, and why does it require high DoF?

A: In-hand manipulation is the ability to shift, rotate, or reorient an object held within the fingers without setting it down on a table (such as rolling a screwdriver or adjusting a key). This requires independent joint movement and finger coordination that only hands with high degrees of freedom (typically 16 to 24 DoF) can provide.

Q: Are tendon-driven hands better than palm-motor hands?

A: Tendon-driven hands move the heavy motors up into the forearm, keeping the hand lightweight, fast, and human-scaled. However, tendons introduce mechanical complexity, cable stretching, and difficult maintenance. Palm-motor hands are more modular and easier to replace, but they are heavier and limited to fewer total degrees of freedom due to internal space constraints.

Explore related platforms and technical profiles in the Bot.to Humanoid Directory or read our direct hardware breakdown: 1X NEO Overview: Why Gearless Tendon Drives Win In-Home Safety.

This divide is embodied in the rivalry between Vancouver-based Sanctuary AI and Silicon Valley-based Figure AI.

Sanctuary AI engineered its Phoenix platform around the belief that the physical world is too chaotic, nuanced, and safety-critical for zero-shot end-to-end neural policies to operate unassisted. Consequently, Sanctuary built Carbon™, a cognitive architecture grounded in high-fidelity, bilateral haptic teleoperation. By placing human operators in the loop via motion-capture rigs and force-feedback gloves, Phoenix captures pristine multi-modal sensory telemetry—including micro-hydraulic pressure gradients and tactile slip forces—using human intelligence to bridge the gap toward incremental automation.

Figure AI adopted the opposing conviction with Figure 02. Backed by custom onboard neural silicon and its proprietary Helix Vision-Language-Action (VLA) foundation model, Figure treats teleoperation not as a permanent operating mode, but as an initial bootstrapping tool to be discarded. Operating autonomously on live automotive assembly lines at BMW Group Plant Spartanburg, Figure 02 processes six RGB-D camera feeds locally at 200 Hz, translating visual tokens directly into joint torques and sub-millimeter part placements without a human supervisor pulling digital puppet strings.

This comprehensive architectural breakdown examines their competing software paradigms, underlying mechatronics, data flywheel efficiencies, and enterprise return on investment (ROI).

Key Architectural Takeaways

• Intelligence Paradigm Divergence: Sanctuary AI relies on a human-in-the-loop teleoperation bridge feeding Carbon AI’s imitation learning pipelines, whereas Figure 02 deploys the autonomous Helix VLA foundation model for direct pixel-to-torque execution.

• Actuation and Manipulation Split: Phoenix packages micro-hydraulic actuators inside its 20-DoF hands to provide sub-millisecond tactile force feedback, while Figure 02 relies on electromechanical cycloidal drives and a self-contained 16-DoF palm-motor hand.

• Data Capture Fidelity: Sanctuary’s bilateral haptic gloves record rich multi-axis pressure, force reflection, and slip telemetry; Figure 02 captures vision-proprioception pairs optimized for scalable neural network training.

• Network Infrastructure Constraints: Phoenix requires ultra-low-latency private 5G or industrial fiber networks for remote pilot safety, while Figure 02 runs entirely disconnected on local dual NVIDIA GPUs.

• Enterprise Validation Vectors: Figure 02 achieved documented daily automotive manufacturing integration with BMW (accumulating over 1,250 production hours), while Phoenix targets logistics kitting with Canadian Tire and component assembly with Magna International.

Quick Specs: Sanctuary AI Phoenix vs. Figure 02 Architecture

The Cognitive Battle: Bilateral Teleoperation vs. End-to-End VLA

The core ideological clash between Sanctuary and Figure centers on how neural networks should interact with the physical world.

Paradigm 1: Sanctuary AI Carbon™ (The Teleoperation-First Data Engine)

• Operational Concept: A human operator wearing a virtual-reality headset, exoskeleton arm trackers, and haptic feedback gloves pilots Phoenix in real time.

• Bilateral Haptic Loop: When Phoenix touches an object, micro-pressure sensors and hydraulic fluid strain gauges measure mechanical resistance and transmit force reflections back into the human pilot’s hands. The operator feels the rigidity of the plastic container or the friction of a threaded bolt.

• Imitation Learning Pipeline: Carbon records every teleoperated movement: 3D visual streams, human joint trajectories, and tactile force vectors. Once hundreds of successful demonstrations are logged across varied lighting and part orientations, Carbon’s machine-learning models train autonomous policies that graduate the task to full automation.

• Explainable Symbolic Logic: Carbon does not rely exclusively on opaque neural weights. It utilizes symbolic reasoning to structure high-level sub-goals (“Align Pin”, “Apply 10 N Torque”), allowing human supervisors to audit robotic decision trees.

↓ (Algorithmic Paradigm Shift)

Paradigm 2: Figure AI Helix VLA (Direct End-to-End Multimodal Transformer)

• Operational Concept: Figure 02 operates autonomously without a remote pilot. A human supervisor gives a high-level verbal or digital directive (“Place the sheet metal fixture onto Pin 4”), and onboard foundation models handle everything else.

• Helix VLA Backbone: Figure’s proprietary Helix model merges visual scene recognition, conversational speech processing, and low-level motor trajectory generation into a single end-to-end neural network architecture.

• Zero-Shot Generalization: Rather than requiring hundreds of human teleoperation runs for every novel item, Helix reasons from its web-scale visual training data to grasp and manipulate unfamiliar parts on first contact, adjusting approach vectors dynamically based on pixel feedback.

• Local 200 Hz Inference: Dual onboard NVIDIA GPUs process camera video locally, generating smooth motor trajectories at 200 Hz without transmitting video packets across external networks.

The engineering trade-off is clear: Sanctuary’s teleoperation pipeline guarantees near-100% operational success on day one because a human brain is piloting the edge cases, but it incurs human labor costs. Figure’s autonomous VLA eliminates remote labor costs and network latency, but requires immense training compute, large fleet deployment data, and sophisticated fail-safe recovery algorithms when unexpected visual edge cases occur.

Hardware Topology: Micro-Hydraulics vs. Electromechanical Cycloidals

A software model is only as effective as the mechanical bandwidth of its actuators. The mechanical differences between Phoenix and Figure 02 explain why each company converged on its chosen software strategy.

Actuator Layout 1: Sanctuary AI Phoenix Micro-Hydraulic Network

• The Forearm & Palm Core: Houses high-pressure micro-fluid manifolds driven by fast-acting proportional piezoelectric and voice-coil valves with sub-millisecond response times.

• Linear Force Density: Miniature hydraulic pistons the diameter of a stylus deliver immense linear thrust directly to the knuckles and thumb base, providing 20 independent degrees of freedom per hand.

• Fluid Damping & Haptic Fidelity: Because hydraulic fluid naturally absorbs mechanical shock and transmits pressure spikes instantaneously without gear lash, it serves as the ultimate sensor for bilateral teleoperation. A remote human operator can literally feel the micro-slip of an object slipping between Phoenix’s fingers.

↓ (Mechatronic Sourcing Shift)

Actuator Layout 2: Figure 02 All-Electric Cycloidal Architecture

• The Modular Rotary Core: Standardizes on frameless brushless DC motors paired with custom cycloidal disc reduction gearboxes across all major structural limbs.

• Clean Industrial Sealing: Completely avoids pressurized mineral oils, manifolds, and hydraulic seals, meeting cleanroom and clean-floor standards required by automotive paint and electronics lines.

• Palm-Integrated Electric Servos: Figure 02’s fourth-generation 16-DoF hand packages tiny brushless servomotors and drive electronics directly inside the palm chassis, driving digits via short synthetic tendons.

• Rotary Precision: Cycloidal gearing provides high positional stiffness, allowing Figure 02 to execute precise, sub-millimeter part insertion tasks repeatedly without positional drift.

While Sanctuary’s micro-hydraulics unlock superior tactile compliance and force sensitivity for teleoperated manipulation, Figure’s sealed electromechanical cycloidal drives are vastly easier to maintain on industrial assembly lines where fluid leaks result in costly line shutdowns.

The Data Flywheel Dilemma: Quality vs. Volume

Humanoid artificial intelligence requires massive amounts of training data. How that data is captured dictates how rapidly an embodied AI model improves.

The Sanctuary Data Engine: High-Density Teleoperation Telemetry

• Data Characteristics: High-dimensional, multi-modal, and human-optimized. Every recorded trajectory contains clean human problem-solving intuition, including how to recover from near-slips using tactile force adjustment.

• Data Cost: High. Capturing 10,000 hours of training data requires 10,000 hours of human operator wages, limiting data velocity to the number of active human pilots on shift.

• Model Convergence: High data quality means downstream imitation learning models require fewer total demonstrations to master complex, contact-rich manual skills.

↓ (Data Acquisition Inversion)

The Figure Data Engine: Autonomous Fleet-Scale Self-Supervision

• Data Characteristics: Autonomous real-world interaction paired with synthetic simulation data.

• Data Cost: Near-zero marginal human cost per hour. As Figure 02 units run autonomously at BMW Spartanburg, they continuously log video frames, joint torques, and success/failure labels directly to internal flash storage.

• Model Convergence: Relies on massive compute clusters to train large-scale VLA transformer models across hundreds of thousands of autonomous iterations, leveraging simulation-to-real (sim-to-real) pipelines to bypass physical teleoperation limits.

Sanctuary AI wagers that high-fidelity tactile teleoperation data produces safer, more nuanced physical models. Figure AI wagers that self-supervised foundation models trained on massive autonomous video datasets will scale faster, mirroring the rapid convergence seen in large language models.

Operational Video Reference: Teleoperation Feedback vs. Autonomous Assembly

The contrast in system execution—human-guided dexterous finesse versus autonomous neural assembly—is visible in active commercial trials:

Sanctuary AI Phoenix Teleoperation and Manipulation:

Watch the platform in pilot testing: Sanctuary AI Phoenix – Inside Commercial Testing

• Key Observation Points:

  • Sub-millisecond bilateral haptic tracking matching remote human operator hand movements.

  • Delicate manipulation of non-rigid, fragile retail goods without structural deformation.

  • Fluid multi-axis wrist articulation supported by forearm micro-hydraulic manifolds.

Figure 02 Autonomous Automotive Assembly:

Watch the platform in BMW production: BMW Group advances Physical AI with Figure in production

• Key Observation Points:

  • Fully autonomous two-handed sheet metal manipulation without a human pilot.

  • Millimeter-accurate insertion of metal locating pins into automotive body fixtures.

  • Continuous 360-degree environmental awareness via onboard RGB-D camera perimeter tracking.

Real-World Commercial Deployment: Magna vs. BMW Spartanburg

The true test of any humanoid architecture is its ability to deliver economic value in live enterprise production environments.

Phase 1: Sanctuary AI Commercial Execution (Magna International & Retail Trials)

• Customer Profile: Tier-1 automotive manufacturing giant Magna International and Canadian retail chain Canadian Tire.

• Application Scope: Automotive sub-assembly kitting, handling flexible wiring harnesses, and sorting thousands of mixed SKUs in warehouse distribution centers.

• Operational Model: Deploys a hybrid workflow where a single remote teleoperator monitors multiple Phoenix units; if an autonomous sub-routine fails or encounters a jam, the human pilot immediately takes control to clear the error.

• Network Requirement: Demands high-bandwidth, ultra-low-latency private wireless infrastructure on customer premises to avoid dangerous control latency spikes.

↓ (Industrial Commercial Contrast)

Phase 2: Figure AI Commercial Execution (BMW Group Plant Spartanburg)

• Customer Profile: Direct automotive OEM integration at BMW’s flagship South Carolina manufacturing facility.

• Application Scope: Sheet metal component placement, chassis sub-assembly kitting, and machine tending within active vehicle body production lines.

• Operational Results: Figure 02 accumulated over 1,250 production hours across ten continuous months, supporting the assembly of more than 30,000 BMW X3 vehicles with zero human teleoperation.

• Network Independence: Completely air-gapped from factory cloud networks; dual onboard NVIDIA GPUs execute all Helix VLA inference locally, ensuring operations continue even if plant Wi-Fi fails.

Engineering Verdict & Field Evaluation

Sanctuary AI Phoenix: Pros & Operational Strengths

• Unmatched Hand Dexterity: 20-DoF micro-hydraulic hands deliver human-equivalent dexterity and tactile sensitivity that electromechanical hands cannot replicate.

• Immediate Zero-Downtime Deployment: Teleoperation guarantees the robot can execute complex, unmapped tasks on day one without waiting months for autonomous neural policy training.

• Bilateral Haptic Data Engine: Captures the richest physical telemetry dataset in the industry, recording true force vectors and tactile micro-slips.

Sanctuary AI Phoenix: Limitations & Engineering Risks

• Teleoperation Labor Overhead: A human-in-the-loop operational model limits gross profit margins and scaling velocity compared to fully autonomous platforms.

• Hydraulic Maintenance on Factory Floors: High-pressure fluid loops introduce seal degradation and oil-leak hazards, complicating deployment in clean automotive assembly zones.

• Low-Latency Network Dependency: Remote teleoperation requires constant low-latency network connectivity; network packet drops create safety risks.

Figure 02: Pros & Operational Strengths

• True Standalone Autonomy: Helix VLA foundation model operates locally with zero human teleoperation, delivering real multi-shift labor replacement.

• Proven Automotive Track Record: 1,250+ documented production hours at BMW Spartanburg confirm enterprise reliability under active assembly line cadences.

• Clean, Sealed Electromechanical Hardware: All-rotary cycloidal drives eliminate fluid maintenance and survive harsh factory dust environments.

Figure 02: Limitations & Engineering Risks

• Edge-Case Generalization Boundaries: When Helix encounters an entirely novel mechanical failure or physically jammed part, autonomous recovery without human intervention remains computationally challenging.

• Lower Hand Kinematic Freedom: 16-DoF palm-motor hands provide less articulation flexibility than Sanctuary’s 20-DoF hydraulic fingers.

The Bot.to Benchmark Verdict:

Figure 02 is the decisive winner in industrial factory automation. By proving that its autonomous Helix VLA stack can operate for ten continuous months on a live BMW assembly line without remote human operators, Figure AI demonstrated that scalable robotics economics require true autonomy. Its sealed electromechanical architecture makes it the gold standard for enterprise manufacturing.

However, Sanctuary AI Phoenix remains the undisputed pioneer in dexterous physical data capture. Its micro-hydraulic hand is a mechanical masterpiece. For hyper-dexterous applications involving flexible wiring, delicate assembly, or complex retail handling where modern autonomous AI still struggles, Sanctuary’s bilateral teleoperation bridge provides a dependable operational solution while building a world-class training dataset for the future.

Frequently Asked Questions (FAQ)

Q: Is Sanctuary AI Phoenix fully autonomous or remote-controlled?

A: Sanctuary AI Phoenix operates using a hybrid model. It utilizes bilateral haptic teleoperation—where remote human operators pilot the robot via VR rigs and force-feedback gloves—to execute complex tasks and gather training data, while its Carbon AI software progressively automates repetitive sub-tasks using imitation learning.

Q: Does Figure 02 require a human operator to control it at BMW?

A: No. Figure 02 operates autonomously on the BMW Spartanburg production line. Its onboard Helix VLA foundation model processes 360-degree camera feeds locally to generate motor trajectories and insert components without real-time human teleoperation.

Q: Why does Sanctuary AI use hydraulics in its hands while Figure 02 uses electric motors?

A: Sanctuary uses micro-hydraulics because miniature fluid pistons deliver immense power density in tiny spaces, enabling 20 degrees of freedom per hand with natural shock absorption and tactile force feedback. Figure 02 uses electric cycloidal drives to eliminate hydraulic fluid leak risks and reduce maintenance on automotive assembly floors.

Q: How does the payload capacity compare between Phoenix and Figure 02?

A: Both platforms have comparable payload capabilities. Sanctuary AI Phoenix is rated for a 25 kg (55 lbs) continuous lift, while Figure 02 is rated for a 20 kg continuous carry with dynamic burst capacities reaching 25 kg.

Explore related platforms and technical profiles in the Bot.to Humanoid Directory or read our direct hardware breakdown: Figure 02 Hardware Architecture: Complete Teardown and Actuator Analysis.