On the Floor: A 2025–26 Field Map of Live Humanoid Deployments in Warehousing, Care and Homes

Executive summary Humanoid robots have moved out of concept labs into real-world pilots, but the footprint remains narrow and pragmatic. Between 2023–mid‑2024 the clearest commercial traction is in logistics and light manufacturing, with healthcare/eldercare and consumer/home trials concentrated in controlled pilots and marketing demonstrations. Vendors position humanoids either as mobile, dexterous augmentations to existing teams (logistics, assembly) or as long‑term strategic bets for care and household assistance; very few vendors have reached scaled, revenue‑generating rollouts. Independent, field‑measured performance data is sparse — most numbers come from vendor disclosures or limited third‑party observers — and outcomes vary by task, environment and integration effort. Where humanoids are actually working today — use‑case profiles Piece‑picking and order consolidation (warehousing) Industry focus: e‑commerce and third‑party logistics. What they do: pick small cartons, tote consolidation, stretch‑goal: mobile replenishment of fast‑moving SKUs. Why deploy: chronic labor shortages for repetitive floor work, peak‑season surges where flexible headcount is costly. Performance (vendor reports): order consolidation throughputs commonly quoted in marketing materials range from ~50–200 picks/hour in highly structured environments; reported error rates are typically Line‑tending and quality inspection (light manufacturing) Industry focus: electronics, small‑parts assembly, automotive sub‑assembly. What they do: tend machines, perform repeatable torque/test steps, carry parts between stations, machine‑vision inspection of assemblies. Why deploy: augment skilled workers on monotonous steps, reduce ergonomic injuries, reconfigure rapidly for short runs. Performance: firms marketing humanoids for co‑bot roles emphasize deterministic cycle times and high uptime when integrated with PLCs, but integration time and software adapters remain the dominant deployment cost. Patient mobility and task assistance (eldercare) Industry focus: long‑term care homes, assisted living pilot wards. What they do: patient transfers (assisted support, not full lifting), fetch/bring tasks, medication delivery between nurses’ stations, and limited social prompts. Why deploy: staffing shortages, infection‑control operations and rising caregiver injury rates. Performance: early trials emphasize safety and comfort rather than throughput; reported uptime is constrained by charging cycles and supervised operating modes. Vendors and care providers report mixed patient acceptance and strict regulatory attention to safe transfer mechanics. Pilot concierge and limited household tasks (consumer/home) Industry focus: affluent early adopters and vendor demonstration homes. What they do: fetch small objects, remote monitoring, controlled choreography/interaction for demos. Why deploy: brand positioning and UX learning rather than mass market replacement. Performance: demonstrations by consumer electronics firms show convincing choreography, but real homes with clutter and unpredictable pets remain a major mobility and perception challenge. Who’s in the field (programs, status and positioning) The field is a mix of logistics pilots, factory co‑bot trials, healthcare proof‑of‑concepts and high‑visibility marketing demonstrations. Below is a snapshot reflecting publicly announced programs and vendor positioning through mid‑2024. Vendor Industry Deployment type Region Notes Agility Robotics (Digit) Logistics Commercial pilots (select DCs) North America, Europe (select) Marketed for last‑mile and fulfillment; vendors cite pilots with logistics operators; focus on pick‑and‑carry consolidation. Apptronik Light manufacturing / industrial co‑bot Pilot trials U.S. Positioned as industrial co‑robot with high payload, integration with PLCs emphasized; trials with manufacturing partners reported. Figure AI, 1X Technologies Assembly, materials handling Lab demos, limited on‑site tests U.S., Asia (labs) Aggressive claims on factory automation; most activity remains testbeds and partner validations. Sanctuary AI, UBTECH Healthcare / eldercare Care‑home pilots and demonstrations Canada, China UBTECH has public eldercare demos in China; Sanctuary focuses on generalist care tasks — both emphasize regulated piloting. Xiaomi, Unitree Consumer / home Showroom demos, limited home pilots China, Global demo tours Primarily marketing and UX validation; Xiaomi showcases appliances integration, Unitree focuses on mobility demos (Unitree is primarily quadruped). Tesla (Optimus) Enterprise demos Lab/enterprise demos — not commercial U.S. Tesla positions Optimus as eventual mass SKU for enterprise and consumer — as of mid‑2024, demonstrations and prototype roadmaps dominate. Customer perspectives and early testimonials Publicly released customer statements emphasize two recurring themes: (1) humanoids reduce specific pain points (repetitive walking, hazardous fetch tasks) and (2) total cost of ownership is driven as much by systems integration as by robot hardware. Logistic operators participating in vendor pilots have described improved flexibility during peak windows but limited gains for complex dynamic picking. Care‑home administrators stress the value of round‑the‑clock monitoring and non‑clinical assistance but highlight patient acceptance, training, and liability as key barriers. Most customer quotes in public releases are high‑level. Independent observers note the difference between a controlled pilot (corridor, labeled bins) and a true, full‑shift deployment in a busy DC or ward. What humanoids displace — and what they still cannot Displacing / augmenting: monotony‑heavy walking/fetch cycles, basic machine tending, repeatable inspection, and low‑variation transfer tasks. Still problematic: dynamic random bin picking, heavy lifting beyond designed payloads, complex social caregiving and unscripted human interaction, robust multi‑level autonomy in cluttered public spaces. Operational constraints observed in the field Battery life: many systems run partial shifts (4–12 hours) with opportunity or tethered charging; continuous 24/7 operation requires redundant fleets and operator oversight. Mobility and clutter: indoor navigation works best in clear aisles; ad‑hoc environments (crowded homes, busy wards) cause frequent human intervention. Systems integration: connecting to WMS, PLCs, and EHRs remains a time‑consuming engineering task — most pilots absorb months of integrator work before useful throughput is reached. Safety & monitoring: vendor safety stacks reduce hard collisions, but human‑in‑the‑loop supervision and geofencing are standard; regulators and insurers remain cautious. Pricing and business models Vendors typically avoid simple list prices. Reported commercial models include pilot-as-a‑service, per‑robot leasing, and outcomes‑based contracts (per‑hour or per‑pick). Early unit cost estimates in public commentary cluster roughly in the low‑to‑mid six‑figure range per unit for fully integrated solutions when accounting for integration, training and support; vendors argue costs will fall with volume. Competitive landscape and takeaways Competition is bifurcated: specialized mobile manipulators and vision‑guided arms remain competitive for constrained tasks; full humanoids compete where legged mobility and human‑like interactions are the differentiator. Agility, Apptronik and Figure push logistics/manufacturing integration; UBTECH and Sanctuary focus on care; Xiaomi and Unitree pursue consumer visibility; Tesla remains a wild card. Success will come where a humanoid’s form factor solves a clear operational problem that prior robots could not — and where integration cost is controlled. Appendix: pilot timelines and vendor claims vs measured outcomes (select) Vendor claim (public) Deployment status Independent/field measurement Digit enables parcel consolidation in live DCs Commercial pilots announced Vendor metrics cite picks/hour; independent verification limited to pilot reports showing benefit in structured lanes only. Apptronik humanoid as industrial co‑bot replacing manual tenders Pilot trials with manufacturers Integration time (weeks–months) dominates; cycle reliability in controlled testbeds good, live floors more variable. UBTECH robots deployed in eldercare for delivery and monitoring Demonstrations and care‑home pilots (China) Positive staff reception for non‑clinical tasks; patient transfer claims remain heavily supervised. Conclusion: through 2024 the most credible, recurring value for humanoids is focused, supervised augmentation in logistics and assembly — tasks that benefit from mobility plus basic dexterity. Healthcare and home use cases show promise but are constrained by regulation, human factors and environmental variability. The next 18–24 months will be defining: vendors that convert pilots into repeatable integrated solutions (with predictable costs for systems integration, charging infrastructure and safety monitoring) will separate commercial pilots from sustained deployments.