Application platform: Humanoid robotics — industrial and service Last verified: 21 August 2026 Update scope: Four documents updated or newly effective July–August 2026; ISO 13482 Edition 2 status check Baseline: Issue #10 Standards Map
Governing standards — current state
No humanoid-specific pack-qualification standard exists. The framework below is entirely adjacent, borrowed from appliance safety, service robotics, general industrial battery standards, and transport regulation. Every pack-level requirement in a humanoid DVP is program-defined. That status did not change this quarter.
Four documents in the adjacent framework did publish new editions, revisions, or reach effective dates between July and August 2026.
| Document | Edition / revision | Date | Scope (one-line) |
|---|---|---|---|
| UL 2595 Ed 3 | Edition 3 | 10 July 2026 | Battery-powered appliance safety supplement |
| UL 3300 Ed 1 rev | August 2026 revision | 14 August 2026 | Safety standard for service robots |
| ISO 18646-5:2026 | Edition 1 | 24 July 2026 | Performance test methods — legged robot locomotion |
| UN 38.3 Rev 8 Am 1 | Amendment 1 | September 2025 | Lithium battery transport testing |
Four adjacent-framework documents updated; no humanoid pack-qualification gap closed. Every DVP row in the gaps table remains program-defined.
Parameter-class cross-reference
Organized by parameter class for readers working a specific DVP row rather than tracking a specific document.
| Parameter class | Documents with Q3 2026 changes | Nature of change |
|---|---|---|
| Power and charging | UL 2595 Ed 3 | Scope expansion (voltage); charging clarification reported but final parameters not public |
| Operating envelope | UL 3300 rev; ISO 18646-5 | Whole-machine moisture conditions; locomotion-test environmental controls |
| Mechanical | UL 3300 rev; ISO 18646-5 | Robot stability, obstacle, locomotion-state standardization; no installed-pack criteria |
| Cycle life | — | No update this cycle |
| Battery abuse / propagation | UL 2595 Ed 3 (reference update only) | No confirmed test-parameter change |
| Environmental protection | UL 3300 rev | Revised moisture/cleaning coverage; no pack IP threshold |
| Transport | UN 38.3 Rev 8 Am 1 | Amendment to settled transport framework |
Nothing in this cycle touched cycle life, charge-rate acceptance criteria, fall impact, bipedal vibration, or human-proximity propagation. The gaps summary at the end of this card is unchanged from the baseline.
UL 2595 Edition 3 — DVP impact
UL 2595 is a supplement, not a standalone pack qualification. It covers lithium-ion battery systems within a battery-powered appliance during storage, discharge, and charge, and it requires an applicable end-product standard to function. Edition 3 leaves that architecture in place.
The change that matters is scope. Maximum rated voltage expanded from 75 Vdc to 250 Vdc (UL 2595 Ed 3, Clause 1; Tier 1 — published standard). Most humanoid pack architectures above 48V nominal exceed 75 Vdc once you account for charge headroom, which put them outside UL 2595's stated scope under Edition 2 and inside it under Edition 3. What moved is applicability, not test content: the expansion determines whether UL 2595's battery-abuse and charging provisions can serve as adjacent-framework evidence for a given platform configuration, and says nothing about whether those provisions themselves changed.
Edition 3 also covers appliances accepting non-rechargeable general-purpose batteries, which Edition 2 excluded where the batteries were user-installed. Per UL's development summary, the edition integrates updated normal-charging requirements for lithium-ion systems and USB-powered products.
The public record does not establish what those updated requirements are. The full Edition 3 redline is not publicly accessible. The UL product page confirms the edition but does not identify changed abuse-test severities, sample counts, charging current or voltage limits, enclosure conditioning parameters, or acceptance criteria relative to Edition 2. No new propagation test, fall test, cycle-life method, or mission C-rate criterion appears in any published summary.
DVP action: Update the adjacent-framework citation from Edition 2 to Edition 3. For platforms above 75 Vdc, note expanded applicability. Do not populate revised charging parameters, abuse-test conditions, or acceptance criteria from the public summary; a controlled Edition 2 / Edition 3 redline is required before assigning changed values to any DVP row. The baseline card's assessment that UL 2595 does not convert appliance evidence into humanoid pack qualification stands.
UL 3300 August 2026 revision — DVP impact
UL 3300 covers service, communication, information, education, and entertainment robots used by or near general consumers. Industrial applications remain outside its scope. An industrial humanoid citing UL 3300 is borrowing from an adjacent framework, same as before the revision.
The published revision summary identifies changes to ambient-light test conditions (§7.2.2.1), robot stability testing including an alternative to the wet-glass ramp (§7.2.4), obstacle testing aligned with ISO 18646-2 (§7.2.5), wet-location definitions and moisture exposure including cleaning and wiping (§§3.5.15, 5.2.3.2, 7.4.10, 8.4.5, 8.11.4), remote-controller operation (§§5.1.6, 8.11.6), elevation-change test thresholds (§7.2.4.1), and manufacturer verification for AI used in safety functions (§8.15).
These are whole-machine behavioral tests. The revision summary identifies no added or revised battery, energy-storage, charging, battery-abuse, or thermal-runaway clause. UL 3300 references UL 2595 Edition 3 and UL 60335-2-29 for battery-powered appliances and battery chargers, but the summary does not indicate those references were newly added by the August 2026 revision.
DVP action: Minimal. The revised moisture-exposure provisions (§7.4.10) supply a whole-robot environmental reference that could inform pack-enclosure test conditions, but the summary gives no battery-enclosure ingress threshold and no pack acceptance criterion. Stability and obstacle tests changed at the machine level; no installed-pack shock pulse, drop height, acceleration spectrum, or post-impact pack criterion is identified. If you are borrowing UL 3300 for environmental-protection or mechanical evidence, update the revision date in your DVP citation. Do not populate new pack-level thresholds from this revision.
ISO 18646-5:2026 — DVP impact
Of the four updates, this one has the most bearing on how a humanoid battery DVP gets written, and it contains no battery parameter at all.
ISO 18646-5:2026 is the first published standard defining performance test methods for legged robot locomotion (Tier 1 — published standard, ISO/TC 299). It covers rated speed (§5), stopping distance (§6), maximum climbing height (§7), stair-climbing capacity (§8), maximum slope angle (§9), dragging force (§10), crossing capacity over gaps (§11), minimum pass-through height (§12), and minimum travel width (§13). ISO states explicitly that these methods evaluate locomotion performance and are not intended for safety verification or validation.
The public preview (Tier 2 — commercial distributor; document identity corroborated by ISO) confirms that the listed measurands are robot-motion and clearance outputs. No energy-consumption, endurance, payload-carrying, battery-capacity, pack current, voltage, power, temperature, SOC, DOD, or degradation test appears in the table of contents. Default environmental range: 10–30°C, 0–80% RH, hard level surface with coefficient of friction 0.75–1.0. The robot must be completely assembled and sufficiently charged per manufacturer instructions.
What the standard does supply is a set of locomotion states with controlled boundary conditions: level-surface speed, stair ascent and descent, slope negotiation at defined angles, obstacle crossing. These are the mechanical loading cases that drive battery duty-cycle definition for a legged platform. The standard will not tell you how to instrument the pack during any of them. It gives you a repeatable way to describe what the robot is doing while you collect electrical data separately, which means a mission cycle defined for pack aging or power-capability testing can anchor its mechanical-load portion to a published reference instead of a program-specific protocol nobody outside your team can reproduce (editorial: this is an inference about DVP utility, not a stated purpose of the standard).
DVP action: ISO 18646-5 does not populate any battery parameter row. Cite it as the source for locomotion-state definitions and boundary conditions when defining a duty cycle, not for battery acceptance criteria. The baseline card's identification of mission cycle life and duty-cycle definition as open gaps is unchanged; this standard supplies a building block for closing that gap at the program level.
UN 38.3 Rev 8 Amendment 1 — DVP impact
UN 38.3 governs lithium battery transport testing and applies to humanoid packs as it does to any lithium-ion battery intended for transport. Transport is the one parameter class on this card with a governing standard, defined tests, and stated acceptance criteria.
Rev 8 Amendment 1 was published in September 2025. The amendment updates provisions within the existing test framework. Transport testing requirements for lithium-ion packs — altitude simulation, thermal cycling, vibration, shock, external short circuit, impact/crush, overcharge, forced discharge — remain structurally unchanged. No new test types, none removed.
DVP action: Update the UN 38.3 citation to Rev 8 Am 1. Verify whether the amendment alters any test parameter, sample size, or acceptance criterion applicable to your specific pack configuration. For most humanoid pack architectures this is a citation update rather than a test-program change, but confirm against the full amendment text before closing the row.
ISO 13482 Edition 2 — status check
ISO/FDIS 13482 remains at Stage 50.00 (FDIS registered for formal approval) as of 21 August 2026. The FDIS was registered 24 July 2025, thirteen months ago. The ISO catalog gives no expected publication quarter. A Japanese government program schedule expected publication in 2026, which is not an ISO commitment.
ISO's public abstract covers service robots in personal and professional/commercial applications, physical human-robot contact, and functional safety. It does not mention batteries, energy storage, or thermal hazards. A commercial preview of the draft table of contents (Tier 2 — corroborated by ISO project record) shows headings for "Hazards due to energy storage and supply" (§4.2), "Hazards related to battery" (§4.2.2), "Uncontrolled release of stored energy" (§4.2.3), and "Power failure or shutdown" (§4.2.4). Battery-related hazards appear in the draft body even though the catalog abstract omits them. The preview does not establish a pack-level test matrix, numerical limits, or final wording.
ISO 13482 Edition 2 is the nearest candidate to eventually address humanoid pack safety inside a service-robot framework. When it publishes, the battery-hazard clauses will need immediate assessment for DVP applicability. Until then the content is not citable as a governing requirement.
Gaps summary — Q3 2026
No Q3 2026 update closed any gap in the table below. This list is unchanged from the Issue #10 baseline card. For broader context on humanoid energy requirements and duty-cycle expectations informing these gaps, see the publication summary's 2026 humanoid energy perspective in this issue.
| Parameter class | Gap | Nearest adjacent reference | Standards activity |
|---|---|---|---|
| Cycle life and calendar life | No humanoid mission-cycle definition, DOD profile, temperature conditions, SOH floor, or life criterion | IEC 62660-1 (automotive cell-level); no pack-level equivalent for robotics | None identified |
| Power capability / charge rate | No humanoid charge-rate acceptance criteria or mission-phase C-rate requirements | Program-specific; baseline card limits reviewed platform evidence to ~0.31C–2.5C (configuration-specific values from disclosed platforms, not a standard) | None identified |
| Mechanical — fall impact | No installed-pack fall pulse, drop height, or post-impact acceptance criterion for bipedal platforms | MIL-STD-810H Method 516.8 (general shock); Figure AI's disclosed 1 m concrete-drop criterion is one OEM's private design precedent, not public authority | None identified |
| Mechanical — bipedal vibration | No vibration PSD or repetition count derived from legged locomotion | MIL-STD-810H Method 514.8 (general vibration); ISO 18646-5:2026 locomotion states could anchor load-case definition but supply no vibration spectrum | None identified |
| Safety — thermal runaway propagation | No human-proximity propagation criterion for installed humanoid packs | IEC 62619 §7.2 (industrial); no humanoid-specific distance, exposure duration, or thermal flux limit | ISO/FDIS 13482 Ed 2 draft includes battery-hazard headings; publication date unknown |
| Safety — abuse tolerance | No humanoid-specific abuse test matrix (nail penetration, crush, overcharge, external short circuit at pack level) | UL 2595 Ed 3 (appliance supplement; final test parameters not publicly confirmed); IEC 62619 §7.2 | None identified |
| Operating envelope — condensation recovery | No pack-level condensation or rapid temperature-transition recovery criterion for humanoid indoor/outdoor transition | UL 3300 §7.4.10 (whole-machine moisture, no pack criterion) | None identified |
| Environmental protection — IP class | No humanoid pack IP rating requirement | UL 3300 §7.4.10 (whole-machine moisture/cleaning); no pack ingress threshold | None identified |
Assessment
The adjacent framework moved this quarter. Legged-robot locomotion now has standardized test methods, whole-machine stability and moisture provisions were revised, and the appliance battery supplement covers the voltage range most humanoid platforms actually operate in. ISO 18646-5 in particular gives DVP authors a better building block than they had last quarter.
The parameter classes that determine whether a humanoid pack is qualified are the same ones that were undefined in Q2. What the battery must survive during a fall, how long it must last under a bipedal duty cycle, how it must behave when it fails within arm's reach of a person — all of that remains the DVP author's problem, defined at the program level and defended with rationale borrowed from adjacent domains. ISO 13482 Edition 2 is the one document with confirmed draft battery-hazard clauses, and it has been at FDIS since last July.
Every row in the gaps table requires a program-defined threshold with documented rationale. The adjacent references listed there are starting points, and should be labeled as such in the DVP.
- ISO 13482 Edition 2: The FDIS has sat at Stage 50.00 for thirteen months with no published ballot or publication-stage entry visible in the ISO project record, and the draft's battery-hazard clauses (§§4.2–4.2.4) will need immediate DVP assessment when the edition finally publishes.
- UL 2595 Edition 3 redline: The ANSI call for comment and UL product page confirm the new edition but do not expose changed abuse-test severities, sample counts, or charging limits, so a controlled Edition 2/Edition 3 comparison remains necessary before populating revised DVP values.
- Humanoid energy-storage constraints: A 2026 peer-reviewed perspective identifies onboard energy storage as a limiting system constraint for humanoids because locomotion and manipulation combine high energy demand, transient power, and close-proximity safety requirements — useful context for the duty-cycle and propagation gaps that remain open.
- Locomotion-state electrical characterization: The open eVTOL aging dataset methodology — mission-shaped loads with distinct power segments across 21,392 cycles — offers a template for how ISO 18646-5's standardized locomotion states could be paired with separate electrical instrumentation to build repeatable humanoid pack-aging protocols.