Shortlist (0)
Draft data. Specs and scores are compiled placeholders, not vendor-verified. How we score.
News · Research

Types of humanoid robots — warehouse, home, research

Source: Robotscomparison — methodology

Digit Figure 03 NEO Unitree G1 +2

Lead strip: Status: Typology primer (not a product) · Availability: n/a · All-in cues: n/a · Scoring: methodology · Shortlist: 2026 evidence shortlist

Job-lane typology — not a sold ranking and not a “best type of humanoid” trophy board.

“Humanoid” is a morphology word. Buyers still need a second label: what job lane the public evidence actually supports. Warehouse tote work, factory sequencing, home pre-orders, research/developer SKUs, and IR roadmaps are different procurement problems even when the robot looks roughly human-shaped. This brief maps those lanes and points to product pages — not a leaderboard.

Robot strips (examples by job lane — not a trophy board)

  • Digit · Agility Robotics · Status: Digit 4 field / Digit 5 announced · Use: warehouse logistics · Evidence: named sites + published hours/totes · → /robots/digit/
  • Figure 03 · Figure AI · Status: Pilot (BMW Spartanburg sequencing) · Use: factory logistics · Evidence: BMW Group partner statement · → /robots/figure-03/
  • NEO · 1X Technologies · Status: Pre-order (consumer-oriented) · Use: home · Evidence: public price path · → /robots/neo/
  • Unitree G1 · Unitree Robotics · Status: Available (research / developer) · Use: lab / EDU · Evidence: public list price · → /robots/unitree-g1/
  • Optimus · Tesla · Status: Research / production-ramp · Use: factory R&D / Academy · Evidence: Tesla IR; external buy page — · → /robots/optimus/

Why “types” should follow jobs, not vibes

Trade press often sorts humanoids by brand fame or demo spectacle. Evidence-first typing asks:

  1. Environment — warehouse aisle, factory hall, dwelling, lab.
  2. Workflow — tote/AMR handoffs, sequencing, chores, research tasks.
  3. Evidence class — field hours, partner pilot, pre-order, list price, IR/roadmap.
  4. Buy surface — RaaS/enterprise, consumer checkout, developer shop, or —.
  5. Generation — do not merge Digit 4/5 or Figure 02/03.

Morphology alone does not tell you which type you are buying into. Full definition checklist: What is a humanoid robot?. Criteria: methodology.

Type 1 — Warehouse / logistics humanoids

Job: material handling in distribution and fulfillment — tote moves, AMR/conveyor handoffs, stacking in human-scale aisles.

Archetype evidence (2026): Digit (Agility Robotics). Agility reports Digit entered regular GXO operations under a multi-year agreement after a late-2023 pilot, with Digit moving more than 100,000 totes at GXO Flowery Branch (Agility: 100k totes). In September 2026 Agility said Digit 4 had logged more than 65,000 operational hours across sites including GXO, Schaeffler, Amazon, and Toyota Motor Manufacturing Canada, while unveiling Digit 5 as the next generation (Digit 5 announcement).

Buyer cues: enterprise / RaaS; public per-unit list —. Digit 4 field evidence ≠ Digit 5 announced (early access framed for H1 2027).

Type 2 — Factory / industrial sequencing humanoids

Job: automotive and plant logistics that sit next to production — sequencing, parts presentation, machine-adjacent material flow — often partner-gated.

Archetype evidence (2026): Figure 03 at BMW Plant Spartanburg. BMW Group states that after a successful Figure 02 body-shop deployment, Figure 03 moves into complex sequencing applications in logistics (BMW Group). Figure describes Hall 52 sequencing under Helix 02 (Figure: F.03 at BMW).

Adjacent industrial path: Atlas (Boston Dynamics) — production-intent electric Atlas with 2026 fleets described as committed to Hyundai’s RMAC and Google DeepMind (Boston Dynamics CES unveil). Strong industrial intent; public retail price —.

Buyer cues: partnership / enterprise. Figure 02 history ≠ Figure 03 current. Factory pilot ≠ consumer home SKU.

Type 3 — Home / consumer-oriented humanoids

Job: household chores and assistance in dwellings; marketing stresses quietness, soft covers, and consumer checkout.

Archetype evidence (2026): NEO (1X Technologies). 1X launched consumer NEO with Early Access $20,000 or $499/month, U.S. deliveries targeted for 2026 and other markets from 2027 (1X NEO, order). NEO Gamma (Feb 2025) was the home-fit development generation that preceded the commercial pre-order stack (Introducing NEO Gamma).

Buyer cues: unusually clear price path. Independent large-scale household fleet proof remains thinner than Digit-class logistics narratives — treat chore autonomy as a claim to verify in real dwellings. Home demos from factory vendors (e.g. Figure Helix kitchen videos) are demo / R&D, not automatic home buyability.

Type 4 — Research / developer / EDU humanoids

Job: labs, universities, secondary development — transparent list prices, SDK/EDU options, thinner industrial fleet receipts.

Archetype evidence (2026): Unitree G1 from about US$13,500 (tax/shipping excluded) on Unitree’s product page (Unitree G1). Entry SKUs such as Unitree R1 publish even lower starting prices for education/entry development.

Buyer cues: high price transparency; do not equate a shoppable biped with warehouse field hours or BMW-class partner sequencing.

Type 5 — Research / IR / internal production-ramp programs

Job: company-internal manufacturing R&D, Academy training data, capacity roadmaps — strategically important, not catalog-buyable until a stable external channel appears.

Archetype evidence (2026): Optimus (Tesla). Public IR materials describe line install and production-ramp language with early builds framed for internal Academy / factory focus; external catalog buy page and stable list price remain — on our Optimus brief (Optimus explained).

Buyer cues: roadmap ≠ purchase order. Score visibility separately from availability.

Quick type matrix (illustrative, not ranked)

Job laneExample platformDominant public evidence (2026)Buy surface
Warehouse / logisticsDigitField hours / named sites / totesEnterprise / RaaS
Factory sequencingFigure 03Partner-stated plant projectEnterprise partnership
Industrial product-intentAtlasCES production-intent + allocated fleetsEnterprise allocation; list —
Home / consumerNEOPre-order / price pathPublic checkout / subscription
Research / developerUnitree G1Public list priceFrom ~US$13.5k (ex tax/ship)
IR / internal rampOptimusTesla IR / Academy framingExternal catalog —

Same morphology class; six different evidence problems. Navigation shortlist (not a rank): 2026 humanoid evidence shortlist.

How to use this typology without trophy language

  1. Pick the job lane first — warehouse, factory, home, lab, or IR.
  2. Match evidence class — field, pilot, pre-order, list price, or roadmap.
  3. Separate generations — Digit 4 ≠ Digit 5; Figure 02 ≠ Figure 03; Gamma ≠ commercial NEO claims without checking.
  4. Ask for the buy surface — if none, write —.
  5. Refuse “best humanoid type” rankings — we do not sell them.

FAQ

Is a warehouse humanoid the same type as a home humanoid?

Morphologically both can be bipedal humanoids. For buyers they are different types by job lane: warehouse evidence is measured in site hours and tote/workflow metrics; home evidence needs residential safety, support, and real dwelling results. Do not copy warehouse KPIs onto home SKUs or vice versa.

Where does Atlas fit?

Industrial / enterprise product-intent — closer to factory sequencing than to consumer home pre-orders. See Atlas and our electric Atlas brief. Public list price: —.

Do you rank types by “best”?

No. This page is a job-lane map plus evidence examples. The evidence shortlist is editorial navigation, not a sold #1 board.

Which type can a private buyer actually order today?

Channel-dependent. Public list/pre-order paths (e.g. G1, NEO) differ from partnership fleets (Digit, Figure 03, Atlas) and IR programs (Optimus). Re-check Status on each /robots/{slug}/.

Sources

Unverified fleet sizes, SLAs, and prices: —.

All news