Humanoid Payload & Reach: Real-World Lifting Limits and Arm Span Comparisons

Understanding Payload and Reach in Humanoid Robots

Payload and reach are the two most frequently cited, yet most frequently misunderstood, specifications in humanoid robotics. Unlike articulated industrial arms that bolt to a fixed base, humanoids must manage a moving center of mass, dynamic balance, and multi-joint coordination. This fundamentally changes how lifting capacity is defined, measured, and validated in practice.

How Payload Is Measured

Manufacturer spec sheets typically rate payload at the end-effector (the gripper or tool flange) under controlled conditions. The rating assumes a static posture, a defined arm extension, and a maximum joint velocity that preserves balance. In reality, payload degrades as the arm extends further from the torso because torque requirements at the shoulder, elbow, and hip joints increase non-linearly. A robot rated for 15 kg at full extension may only safely handle 8 to 10 kg when operating near its kinematic limits, or when accelerating the load during pick-and-place cycles.

Independent validation requires factory videos, joint torque telemetry, or third-party load-cell testing. We prioritize claims backed by shipping hardware telemetry over render-based demonstrations. When a manufacturer publishes a spec sheet, the payload rating should explicitly state the posture (arm fully extended vs. close to torso), the speed profile, and whether the rating includes the weight of the gripper itself. Without these qualifiers, the number is functionally meaningless for integration planning.

Reach Envelopes and Workcell Constraints

Reach describes the maximum vertical and horizontal extension a humanoid can achieve while maintaining stability. Most commercial humanoids list a maximum reach between 1.5 meters and 1.7 meters. However, practical reach is constrained by three factors:

• Center of mass shift: Extending the arm too far horizontally requires counterweighting via hip rotation or step adjustment, which slows cycle time.
• Joint hard limits: Mechanical stops and software safety envelopes typically cap extension at 90 to 95 percent of the theoretical kinematic maximum.
• Task geometry: Workcells designed for fixed-base arms often assume a static footprint. Humanoids require a swing radius and a safety perimeter, effectively reducing usable reach by 15 to 20 percent in tight layouts.

When evaluating reach, treat the published number as a theoretical envelope, not a usable workcell boundary. Integration planning should account for dynamic balance recovery, gripper compliance, and the robot's ability to reposition its base without losing the load.

Shipping Hardware: Verified Payload and Reach Specs

We grade claims by shipping hardware first, pilot deployments second, and announcements last. The following specifications are drawn from manufacturer spec sheets, factory telemetry, and verified on-stage demonstrations where available.

Figure 02

Figure AI's 02 unit is currently shipping to enterprise pilot sites. The published spec sheet rates a 15 kg payload at full arm extension and a maximum reach of approximately 1.7 meters. The design uses a combination of series elastic actuators and high-ratio harmonic drives to maintain joint stiffness under load. Independent factory footage shows stable pick-and-place cycles with 12 to 15 kg loads at moderate speeds, with payload dropping to roughly 10 kg when operating at maximum extension and acceleration. The 02 is not yet available through direct India distributors, but pilot units have been deployed in North American and European logistics facilities.

Unitree G1 and H1

Unitree Robotics ships the G1 as a lightweight research and development platform. The G1 is rated for a 5 kg payload at full extension, with a reach of approximately 1.55 meters. The design prioritizes low inertia and high bandwidth, making it suitable for rapid motion rather than heavy lifting. The H1, positioned as a heavier platform, carries a 15 kg payload rating and a reach of roughly 1.65 meters. Unitree publishes joint torque curves and power consumption data alongside payload ratings, which aligns with independent teardown reports. Both units are available for direct purchase, with the G1 shipping at a significantly lower price point than heavy-duty industrial alternatives.

Tesla Optimus (Gen 2/3 Updates)

Tesla has demonstrated the Optimus platform handling 20 kg at full extension during public updates. These demonstrations are informative but lack the standardized telemetry required for industrial integration. The published target payload remains 20 kg, with a reach of approximately 1.7 meters. However, factory deployment timelines and software maturity around force control and grip modulation remain in active development. Until Tesla publishes standardized spec sheets with verified cycle-time data under load, the 20 kg rating should be treated as a design target rather than a guaranteed operational limit.

Apptronik Apollo

Apptronik's Apollo is built for logistics and warehouse integration. The published payload is 15 kg at full extension, with a reach of roughly 1.6 meters. Apollo's architecture emphasizes deterministic control loops and gripper force feedback, which aligns with its target use case of repetitive box handling. The unit ships with documented joint limits and safety envelopes, making it one of the more transparent platforms for payload planning. Pilot deployments have focused on US and European distribution centers.

Pilot Deployments: What Factory Floors Actually Report

Pilot data reveals the gap between lab specs and shop-floor reality. Across multiple logistics and assembly pilots, the following patterns consistently emerge:

• Payload derating under motion: Robots that handle 15 kg statically often require 20 to 30 percent derating when accelerating or decelerating the load.
• Gripper compliance matters: Standard parallel-jaw grippers lose effective payload by 10 to 15 percent when handling irregular shapes or requiring adjustable grip force.
• Thermal limits: Continuous heavy lifting triggers joint thermal protection, forcing duty cycle reductions that impact throughput calculations.

These observations are not theoretical. They appear in pilot reports from third-party integrators who track cycle time, joint temperature, and grip slippage. When evaluating a humanoid for heavy work, request thermal derating curves and dynamic payload tables, not just static spec sheet numbers.

Announcements and Concept Claims: Reading the Fine Print

The humanoid sector is saturated with rendered concepts and keynote promises. We grade announcements last for a reason: they lack telemetry, compliance testing, and production-grade actuators. When a company announces a 25 kg payload or 2-meter reach, check for:

• Published spec sheets with posture qualifiers
• Independent load-cell validation or factory telemetry
• Shipping hardware with documented duty cycles

If only render animations or keynote slides exist, treat the payload and reach numbers as directional targets, not integration-ready specifications.

India Availability and Approximate INR Pricing

Humanoid robots are not yet mass-distributed in India. Direct sales are limited to pilot programs, research partnerships, and select enterprise deployments. Import routes typically involve authorized distributors or direct procurement from North American and Chinese manufacturers. Landed costs in India are heavily influenced by customs duties, software licensing, and after-sales support packages.

Approximate landed cost estimates for India (clearly flagged as estimates) range from ₹1.8 crore to ₹3.2 crore per unit, depending on configuration, gripper options, software stack, and import duties. Research platforms like the Unitree G1 sit at the lower end, while enterprise logistics models like Apollo or Figure 02 approach the higher end. These figures include base hardware, standard grippers, and initial software licensing. They do not include facility integration, safety fencing, or ongoing maintenance contracts.

For Indian integrators, we recommend verifying distributor status, checking GST and import duty classifications, and requesting pilot access before committing to procurement. Humanoid payload and reach capabilities are still maturing, and direct deployment without pilot validation carries integration risk.

How to Evaluate Payload Claims Before Buying

When reviewing humanoid specifications for industrial use, follow this checklist:

• Verify the posture qualifier: Payload at full extension is not the same as payload at 45 degrees.
• Request dynamic load tables: Static ratings ignore acceleration, deceleration, and grip modulation.
• Check thermal derating curves: Continuous heavy lifting triggers joint protection that reduces throughput.
• Confirm gripper weight inclusion: Some manufacturers exclude the gripper from the payload rating.
• Ask for pilot telemetry: Factory footage and third-party integration reports are more reliable than keynote demonstrations.

Payload and reach are not marketing numbers. They are engineering constraints that dictate cycle time, safety margins, and workcell design. Treat them as such.

References

• Figure AI Specifications and Technical Data. https://www.figure.ai/specifications
• Unitree Robotics G1 and H1 Technical Documentation. https://www.unitree.com/g1
• Apptronik Apollo Product Overview and Integration Guide. https://www.apptronik.com/apollo
• Tesla Optimus Platform Updates and Technical Briefings. https://www.tesla.com/optimus
• IEEE Robotics and Automation Magazine: Humanoid Locomotion and Load Handling Constraints. https://ieeexplore.ieee.org
• International Federation of Robotics: Humanoid Integration and Payload Validation Guidelines. https://www.ifr.org