Because simulation still struggles with micro-scale contact dynamics, soft-tissue deformations, and oily surface friction, high-precision human teleoperation remains the gold standard for collecting real-world robotic demonstration datasets.

Yet, operating an anthropomorphic bipedal humanoid with 30+ degrees of freedom is fundamentally different from flying an aerial drone or driving an industrial mobile base.

The teleoperation workstation must bridge two radically different physical systems: the biomechanics of the human body and the non-biological kinematic constraints of an electromechanical robot.

Human-in-the-loop robotic teleoperation rig. Source: CFOTO / CFOTO/Future Publishing via Getty Images

A human operator possesses flexible spherical joints, variable skin compliance, and continuous subconscious vestibular balance adjustments.

The robot, by contrast, operates with rigid cycloidal gearboxes, discrete joint limits, fixed link lengths, and whole-body center-of-mass constraints.

If an operator moves their hand with an acceleration that exceeds the robot’s actuator torque limits, the robot lags, tracking error accumulates, and the collected training trajectory becomes kinematically invalid.

Robotics companies navigate this gap using two primary workstation paradigms: Immersive Spatial VR Rigs (Apple Vision Pro, Meta Quest Pro) and Direct-Drive Force-Feedback Exoskeleton Rigs.

This technical breakdown examines the hardware architectures, bilateral haptic feedback pipelines, real-time kinematic retargeting algorithms, latency budgets, and data-logging schemas that power modern physical AI training workstations.

Key Architectural Takeaways

• The Teleoperation Paradigm Split: Spatial VR rigs provide high operator comfort, rapid setup, and low hardware cost ($1,500–$4,000), but lack physical contact resistance; bilateral master-slave exoskeleton rigs ($50,000–$150,000) deliver direct mechanical force-reflection, which is necessary for sub-millimeter force-guided insertion tasks.

• The Dynamic Latency Budget: To prevent operator motion sickness and eliminate unstable command oscillations, end-to-end glass-to-glass and hand-to-joint latency must remain bounded strictly under 50 milliseconds.

• Kinematic Retargeting via Null-Space Optimization: Human arm joint configurations cannot map 1:1 onto humanoid linkages; real-time solvers prioritize 6-DoF end-effector tracking while projecting human elbow and shoulder angles into the null-space of the robot manipulator Jacobian.

• Bilateral Haptic Transparency: High-performance exoskeleton rigs utilize transparent quasi-direct drive (QDD) actuators on the master side to reflect remote contact wrenches (Fext​) to the operator’s hands without adding virtual inertia.

• Multimodal Data Packaging (HDF5 / MCAP): A single 8-hour teleoperation shift produces 200 to 500 gigabytes of synchronized data per station, recording multi-view RGB-D video, high-rate joint states, tactile pressure maps, and operator gaze vectors time-aligned to $<1\text{ ms}$.

Quick Specs: Spatial VR Suites vs. Direct-Drive Exoskeleton Workstations

The Workstation Architecture: Hardware Topology

A production teleoperation station is an integrated physical-digital cockpit engineered to maintain continuous operational immersion:

Real-Time Kinematic Retargeting: Solving the Cross-Embodiment Gap

A human arm cannot be mapped directly onto a humanoid robot through simple 1:1 joint angle copying.

A human shoulder is a complex biological ball-and-socket mechanism with a sliding scapula; a humanoid robot’s shoulder typically consists of two or three intersecting orthogonal revolute motors.

Furthermore, human limb lengths and range-of-motion limits vary significantly across operators.

Teleoperation systems resolve this using Real-Time Optimization-Based Kinematic Retargeting:

Real-Time Retargeting Execution Loop:
[Human Hand Pose via VR Controller / Exoskeleton Encoders (1,000 Hz)]
                                ↓
[Kinematic Workspace Scaling: Adapts Human Proportions to Robot Arm Length]
                                ↓
[Hierarchical Quadratic Program (QP) Solver: Solves Primary 6-DoF End-Effector Tracking]
                                ↓
[Null-Space Projection: Solves Secondary Elbow Swivel to Match Human Posture]
                                ↓
[Safety Clamping: Evaluates Self-Collision Meshes and Actuator Limits]
                                ↓
[Target Commands Dispatched to Physical Robot Inverters Over Low-Latency Bus]

1. Primary Task: 6-DoF End-Effector Control

The solver prioritizes keeping the robot’s gripper pose matched to the human operator’s hand in Cartesian space ($X,Y,Z,\text{Roll},\text{Pitch},\text{Yaw}$).

If the operator moves their hand forward 10 centimeters, the robot’s gripper moves forward exactly 10 centimeters, regardless of the intermediate joint angles required.

2. Secondary Task: Null-Space Postural Optimization

A 7-DoF humanoid arm has a redundant degree of freedom: it can rotate its elbow outward or inward without moving the hand.

The retargeting engine projects the operator’s measured biological elbow angle into the null-space of the Jacobian matrix.

This allows the robot to mirror natural human arm postures, ensuring that the arm maneuvers around workcell obstacles rather than flailing into side walls.

Bilateral Force Feedback: Letting the Operator “Feel” Remote Contact

When an operator relies entirely on visual cues through a VR headset, they cannot detect when a component makes contact.

A human operator attempting to slide a peg into a hole using visual feedback alone will inadvertently push too hard, bending pins or stalling robot joints before noticing the alignment error on the screen.

Bilateral Teleoperation closes this loop by projecting physical forces bi-directionally:

The Passivity and Stability Challenge:

Transmitting force and velocity signals across a network connection introduces phase lag.

In bilateral control, phase lag turns a closed feedback loop into an active energy generator: if an operator hits a hard wall, network latency can cause the master exoskeleton to push back violently after the operator has already stopped, creating dangerous mechanical resonance.

To guarantee operator safety, production workstations implement Time-Domain Passivity Controllers (TDPC) or Wave Variable Formulations:

• Continuously monitors the net energy flux flowing across the communication channel:

  E(t)=∫0t​(Fmaster​⋅vmaster​−Fslave​⋅vslave​)dt

• If latency spikes cause the system to generate virtual energy ($E(t)<0$), an adaptive software damper instantly dissipates the excess force, preserving physical stability and protecting the operator’s joints.

End-to-End Latency Budgets: The 50 ms Reality Ceiling

In teleoperation, latency is not simply an inconvenience; it directly degrades the quality of the collected training data.

If total glass-to-glass latency exceeds 60 milliseconds, human operators experience sensory disconnect: their motor cortex commands a movement, but the visual confirmation arrives with a noticeable lag.

Operators begin to over-correct, producing unnatural, hesitant, “stop-and-wait” movement trajectories that degrade imitation learning models.

The teleoperation loop operates within a strict 50-millisecond latency envelope:

Multimodal Data Logging: The Structure of an AI Training Dataset

The ultimate output of a teleoperation workstation is not the physical part assembled on the table; it is the multimodal dataset logged to disk.

Every movement, visual frame, and tactile interaction is serialized into structured, high-throughput container formats (such as HDF5, Zarr, or ROS 2 MCAP files) for training Vision-Language-Action models:

Logged Demonstration Frame Schema (Sampled @ 50 Hz):
{
  "timestamp_utc_epoch_ns": 1726308888123456789,
  "episode_id": "ep_assembly_m6_bolt_0421",
  "task_language_instruction": "Align the bracket and torque the M6 bolt into fixture B",
  "observations": {
    "camera_head_rgb": [1920, 1080, 3],        // Compressed H.265 frame
    "camera_wrist_left_rgb": [1280, 800, 3],   // Grasp contact view
    "camera_wrist_right_rgb": [1280, 800, 3],  // Tool orientation view
    "joint_positions": [32],                   // Radians (Float32)
    "joint_velocities": [32],                  // Rad/s (Float32)
    "joint_measured_torques": [32],            // Nm (Float32)
    "wrist_fts_wrench_left": [6],              // [Fx, Fy, Fz, Tx, Ty, Tz]
    "wrist_fts_wrench_right": [6],             // [Fx, Fy, Fz, Tx, Ty, Tz]
    "tactile_skin_taxels_left": [128],         // Pressure distribution
    "tactile_skin_taxels_right": [128]
  },
  "actions": {
    "commanded_joint_positions": [32],         // Target sent to inverters
    "commanded_cartesian_pose_left": [7],      // [X, Y, Z, Qx, Qy, Qz, Qw]
    "commanded_cartesian_pose_right": [7],
    "commanded_gripper_effort": [2]
  }
}

Quality Control and “Golden Run” Pruning:

Not all teleoperation data is suitable for training.

If an operator sneezes, hesitates, misses a grasp, or drops a component, logging that raw trajectory teaches the downstream neural policy to hesitate and drop parts.

Workstations integrate automated quality scoring pipelines:

• Jerk Cost Auditing: Calculates the integral of squared jerk (∫∥x…​∥2dt). Erratic, trembling human hand trajectories are flagged for human review.

• Force-Limit Trajectory Pruning: Any demonstration where contact wrenches exceed critical safety limits (e.g., side loads $>60\text{ N}$) is automatically isolated.

• Success Tagging: A supervisor station verifies successful electrical engagement before the episode is promoted into the golden training pool.

Engineering Verdict & Field Evaluation

Spatial VR Workstations: Pros & Strategic Strengths

• Rapid Scale-Out: Low capital cost allows robotics companies to deploy dozens of data-collection booths across multiple facilities quickly.

• High Operator Ergonomics: Lightweight spatial headsets (such as Apple Vision Pro or Quest Pro) reduce physical strain, enabling longer continuous recording shifts.

• Broad Workspace Agility: Allows unconstrained tracking across large movement volumes without mechanical arm collision or singularity limits.

Direct-Drive Exoskeleton Workstations: Pros & Strategic Strengths

• Unmatched Force-Guided Accuracy: Force-reflecting linkages are essential for teaching robots high-precision mechanical insertions, screw threading, and tactile seatings.

• True Kinematic Determinism: Absolute optical joint encoders track joint angles without the drift, occlusions, or lighting dropouts common in optical VR controllers.

• Elimination of “Phantom” Collisions: Direct physical feedback stops the human operator from commanding impossible kinematic trajectories into rigid machinery.

The Bot.to Benchmark Verdict:

The ideal industrial teleoperation strategy is not a single workstation design; it is a tiered, specialized data collection architecture.

For coarse material handling, pallet kitting, tote transfer, and navigation, immersive spatial VR headsets running low-latency H.265 pipelines deliver the highest data throughput at the lowest cost per trajectory minute.

However, for high-precision assembly lines involving micro-scale tolerances, force-sensitive electronic components, and delicate mechanical insertions, bilateral force-reflecting exoskeleton rigs remain indispensable.

By investing in high-rate bilateral haptics, low-latency glass-to-joint transport, and automated kinematic retargeting, robotics organizations ensure their foundation models train on clean, physically viable demonstrations that translate into reliable industrial performance.

Frequently Asked Questions (FAQ)

Q: Why can’t engineers just use computer vision to teach robots without human teleoperation?

A: Computer vision alone cannot show a robot the physical contact forces, delicate tactile adjustments, and subtle compliance required to assemble parts. When fingers grasp a part, the camera’s view is physically blocked by the hand itself (visual occlusion). Teleoperation allows an expert human to directly demonstrate how to feel for alignment, regulate grip pressure, and recover from micro-slips in the real world.

Q: What is “bilateral force feedback” in robotic teleoperation?

A: Bilateral force feedback is a two-way control system. As the human operator moves the master rig to control the robot, force sensors on the robot measure contact resistance against objects in the real world and transmit those forces back to motorized actuators on the master rig. This allows the human operator to physically feel the weight, stiffness, and surface friction of whatever the robot is touching.

Q: What is kinematic retargeting, and why is it necessary?

A: Kinematic retargeting is the mathematical process of translating a human operator’s limb motions into joint commands that a robot can execute. Because human joints (like the sliding human shoulder joint) do not match the rigid mechanical pivots of a robot arm, algorithms must continuously scale limb lengths, avoid the robot’s physical joint limits, and prevent the robot from colliding with itself while preserving the hand’s target path.

Q: Why is latency so critical during robot teleoperation?

A: If the delay between the operator moving their hand and seeing the robot react exceeds 50 milliseconds, the human brain perceives a disorienting lag. Operators begin to over-correct their motions, causing the robot to shake, bang into fixtures, or drop parts. Low-latency streaming (sub-50 ms) is necessary to ensure the captured movement data is smooth, fluid, and natural enough to train AI models.

Explore related platforms and technical profiles in the Bot.to Humanoid Directory or read our direct hardware breakdown: Multi-Camera Spatial SLAM: How Humanoids Map Dynamic Factory Environments Without LiDAR.

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.