Humanoids in Logistics: Where Figure, Apptronik and Agility Are Actually Deployed

The Shipping Hardware Tier: What Has Actually Left the Factory

Humanoid logistics claims must be graded by physical hardware delivery first. Until units cross the threshold from prototype benches to operational environments, deployment claims remain in the announcement category. As of the current reporting cycle, three manufacturers have shipped hardware to logistics and industrial partners for field evaluation.

Figure: BMW, UPS, and Amazon Integration Hardware

Figure has shipped its Figure 02 and early Figure 03 units to multiple industrial sites. The hardware was deployed to BMW Group's Spartanburg plant for automotive logistics testing, where it handled parts transport and assembly line support under supervised conditions. Figure also delivered hardware to UPS facilities for package handling trials and to Amazon fulfillment centers for cargo sorting evaluation. Shipping hardware to these sites confirms mechanical readiness and basic mobility, but does not equate to autonomous commercial deployment. The units operate with remote oversight, periodic software updates, and fixed charging infrastructure.

Apptronik Apollo: Walmart and FedEx Pilot Hardware

Apptronik has shipped Apollo units to Walmart distribution centers and FedEx facilities. The hardware is configured for inventory scanning, package sorting, and facility navigation. Apollo's mechanical design emphasizes torque-controlled joints and a standardized end-effector platform, which simplifies integration with existing warehouse management systems. However, the shipped hardware remains in pilot mode. Battery life, dexterity under variable load, and navigation in high-traffic aisles are measured against operational baselines rather than commercial service targets.

Agility Digit: Amazon's Cargo and Sorting Deployment

Agility Robotics has shipped Digit units to Amazon fulfillment centers. The deployment focuses on cargo loading, unloading, and sorting operations. Digit's hydraulic-electric hybrid actuation and stereo vision stack are evaluated for repetitive material handling. Amazon's internal reporting confirms hardware delivery and ongoing field testing, but the units are restricted to controlled zones with manual intervention protocols. The hardware meets baseline durability requirements for warehouse environments, yet software latency and environmental perception limits prevent unsupervised scaling at this stage.

Pilot Deployments and Operational Realities

Pilot deployments reveal the actual task scope, failure modes, and integration friction. Humanoid logistics pilots are graded second, and the data shows narrow applicability rather than broad warehouse transformation.

Task Scope and Limitations in Warehouse Environments

• Package sorting and cargo handling remain the primary pilot tasks. Humanoid form factors are tested for compatibility with existing conveyor systems and manual workstations.
• Dexterity constraints limit manipulation to standardized box sizes and pre-positioned pallets. Irregular shapes, flexible packaging, and dynamic stacking require vision recalibration and force-torque tuning.
• Battery endurance averages 4 to 6 hours under continuous operation. Swapping stations and charging infrastructure dictate shift planning more than software updates.
• Navigational reliability depends on pre-mapped facility layouts. Dynamic obstacles, temporary shelving, and cross-traffic reduce autonomy scores during peak hours.

Software Stacks and Perception Constraints

Perception and control layers are the primary bottleneck. Manufacturers report improved foundation models for object detection and grasp planning, but real-world variance in lighting, surface reflectivity, and packaging wear introduces error rates that require human verification. Edge computing on humanoid platforms remains power-constrained, forcing reliance on cloud inference for complex scene understanding. Latency spikes during network congestion have caused stoppages in pilot trials, reinforcing that humanoid logistics remains a supervised integration rather than an autonomous workflow.

Announcements vs. Commercial Readiness

Announcements are graded last. Press releases frequently cite partnerships with major retailers and logistics firms, but the commercial reality is measured by units operating without manual override. Current deployments fall short of unsupervised commercial service. Fixed robotics, autonomous mobile robots, and vision-guided arms continue to dominate warehouse automation due to higher throughput, lower maintenance costs, and proven reliability. Humanoids are evaluated for flexibility in mixed-use environments, but flexibility introduces mechanical wear, calibration drift, and higher per-unit operating costs. Until software stacks achieve consistent zero-touch operation and hardware meets 24/7 duty cycles, logistics announcements remain forward-looking rather than operational.

India Availability and Landed Cost Estimates

India availability for commercial humanoid logistics hardware is currently zero. No manufacturer has announced formal sales, local assembly, or authorized service partnerships in the Indian market. Pilots remain confined to North America and select European facilities.

Supply Chain, Duty Structures, and Service Infrastructure

Importing humanoid logistics hardware into India requires navigating multiple cost layers. The estimated landed cost per unit, including shipping, insurance, customs duties, GST, and initial service contracts, ranges between ₹3.8 crore and ₹4.2 crore. This estimate is flagged as a preliminary calculation based on current import tariffs, freight rates, and the absence of local assembly or component sourcing. Indian warehouses will likely adopt tiered automation first: AMRs for transport, fixed arms for pick-and-place, and vision systems for quality control. Humanoid integration will follow only after software reliability improves, service networks expand, and duty structures for robotics components are revised. Until then, logistics operators in India should treat humanoid deployments as experimental rather than operational.

References

• Figure AI. "Figure Partners with BMW Group." https://www.figure.ai/blog/bmw-figure
• Figure AI. "Figure Partners with UPS." https://www.figure.ai/blog/figure-partners-with-ups
• Figure AI. "Figure Partners with Amazon." https://www.figure.ai/blog/figure-partners-with-amazon
• Apptronik. "Apptronik and Walmart Partner on Apollo Humanoid Pilots." https://www.apptronix.com/news/apptronik-walmart
• Apptronik. "Apptronik and FedEx Collaborate on Logistics Pilots." https://www.apptronix.com/news/apptronik-fedex
• Agility Robotics. "Digit Deployed at Amazon Fulfillment Centers." https://www.agilityrobotics.com/news/agility-robotics-digit-deployed-at-amazon
• Agility Robotics. "Digit Technical Specifications." https://www.agilityrobotics.com/digit
• Apptronik. "Apollo Product Specifications." https://www.apptronix.com/apollo