Payload & Reach: Verified Lifting Capacities and Arm Extents in Humanoid Robots

Defining Payload and Reach in Humanoid Robotics

Payload and reach are the two most frequently cited, yet most frequently misunderstood, specifications in humanoid robotics. Payload refers to the maximum mass a robot can safely hold statically or manipulate dynamically, while reach defines the furthest point the end-effector can extend from the robot's base or joint origin. In industrial and logistics contexts, these numbers dictate whether a humanoid can replace a human worker on a specific task, not merely demonstrate mobility in a lab.

At RobotWale, we grade claims strictly by deployment maturity: shipping hardware first, verified pilot deployments second, and public announcements last. Static lift capacity is only one component of utility. Dynamic payload, speed, battery depletion under load, wrist torque, and tooling dependency determine whether a spec translates to operational value.

Grading the Market: Shipping Hardware, Pilots, and Announcements

Shipping Hardware (Verified)

• Figure 02 (Figure AI): Rated for 20 kg payload at full reach. The unit is in active pilot deployments across US and European manufacturing sites. Reach is approximately 1,850 mm from the hip joint. The specification sheet emphasizes torque-controlled actuators and force-torque sensors in the wrists, which directly impact dynamic handling rather than static hold.
• Apptronik Apollo: Rated for 20 kg payload with a reach of 1,830 mm. Apollo has shipped to pilot customers and is deployed in controlled warehouse and manufacturing environments. The design prioritizes safety margins and certified torque limits over maximum lift capacity.
• Unitree G1 & H1: The G1 specifies 2 kg payload at the wrist with a reach of roughly 1,400 mm, optimized for agility and cost. The H1, designed for heavier industrial tasks, lists 20 kg payload capacity. Both units have shipped to research and industrial pilot partners. Payload ratings assume rigid tooling and steady-state operation.
• Agibot B2: Official specifications list 20 kg payload with a reach of 1,900 mm. Agibot has begun shipments to pilot customers in China and is expanding distribution. The B2 uses high-torque series elastic actuators to manage dynamic loads and maintain joint stability under weight.

Pilot Deployments (Field-Tested)

• Tesla Optimus Gen 2: Tesla has demonstrated payload handling in factory pilot programs, citing 20 kg capacity with a reach of 1,800 mm. The unit has undergone extensive line-testing at Tesla facilities. Dynamic payload in motion remains lower than static hold, particularly when accelerating or decelerating heavy loads.
• BMW & Geely Pilots: Both automakers have deployed multiple humanoid platforms on assembly lines. Field data consistently shows that effective dynamic payload drops by 15–25% when handling irregular objects, using suction cups, or operating at speeds above 0.5 m/s. Reach remains consistent, but wrist torque limits dictate safe handling envelopes.

Announcement-Stage Claims (Unverified)

• Several startups and legacy manufacturers have announced units claiming 30 kg or 40 kg payload with reach exceeding 2,000 mm. These figures are typically static hold ratings under ideal conditions, with no published data on battery impact, cycle life, or tooling requirements. Until hardware ships and independent testing confirms the numbers, these claims remain marketing projections rather than operational specifications.

Static Hold vs. Dynamic Handling: What the Numbers Actually Mean

A 20 kg static payload rating does not translate to 20 kg of usable lifting capacity in a production environment. Dynamic payload depends on joint acceleration limits, gear backlash, battery voltage sag, and the type of end-effector. When a humanoid lifts a box, the motors must overcome inertia, not just gravity. Wrist torque limits typically cap dynamic payload at 10–15 kg for safe, repeatable operation.

Reach specifications are measured from the hip or base joint to the tool center point. Longer reach increases moment arm stress on the shoulder and elbow joints, reducing safe payload capacity. Robots with shorter reach but higher joint torque often outperform longer-reach units in dense assembly or packaging tasks.

India Availability and Approximate Landed Pricing

Humanoid robots remain a niche import category in India. Most units are sourced through authorized distributors, research partnerships, or direct pilot contracts. Customs duties, GST, and compliance certifications affect landed costs significantly. The following are approximate landed cost estimates, clearly flagged as non-binding market estimates based on current import patterns and distributor quotes:

• Figure 02: Approximately ₹2.8–3.2 crore per unit (landed). Requires pilot contracts and technical onboarding.
• Apptronik Apollo: Approximately ₹2.5–2.9 crore per unit (landed). Distributed through industrial automation partners in Mumbai and Bangalore.
• Unitree G1: Approximately ₹1.1–1.4 crore per unit (landed). Available through robotics integrators in Delhi NCR and Hyderabad.
• Unitree H1: Approximately ₹1.6–1.9 crore per unit (landed). Focused on research institutions and heavy-automation pilots.
• Agibot B2: Approximately ₹1.4–1.7 crore per unit (landed). Distribution expanding through Chennai and Pune industrial corridors.

Indian procurement typically involves testing periods, safety certifications, and integration with existing PLC or MES systems. Landed costs exclude installation, training, and maintenance contracts, which can add 15–20% to the total cost of ownership.

Evaluation Checklist for Procurement

When evaluating payload and reach claims for deployment, prioritize verified data over marketing materials. Use this checklist to assess real-world utility:

• Static vs. Dynamic Payload: Request torque curves and acceleration limits under load. Dynamic payload is the operative metric for line deployment.
• Reach Measurement Standard: Confirm whether reach is measured from the hip, base, or shoulder joint. Longer reach does not equal better utility if joint stress reduces payload capacity.
• Tooling Dependency: Identify whether the payload rating assumes custom end-effectors, suction pads, or grippers. Standard grippers reduce effective payload by 20–30%.
• Battery Impact Under Load: Ask for cycle life and voltage sag data. Heavy payloads drain batteries faster and require more frequent charging or swapping.
• Safety Certifications: Verify CE, ISO, or Indian BIS compliance. Payload ratings without safety margins are operationally irrelevant.
• Pilot Data: Prioritize units with published field reports. Real-world handling data reveals limitations that spec sheets obscure.

References

1. Figure AI, "Figure 02 Technical Specifications," https://www.figure.ai/specs
2. Apptronik, "Apollo Product Specifications," https://www.apptronik.com/apollo
3. Unitree Robotics, "H1 & G1 Technical Data Sheets," https://www.unitree.com/products/h1
4. Tesla, "Optimus Gen 2 Specifications (AI Day 2023)," https://www.tesla.com/AI
5. Agibot, "B2 Humanoid Robot Official Press Release," https://www.agibot.com/en/product/b2
6. Reuters, "BMW and Geely Deploy Humanoid Robots in Manufacturing Pilots," https://www.reuters.com/technology/