The Standards Map — Humanoid Robotics Pack Qualification

Application platform: Humanoid Robotics Last verified: 14 August 2026

Governing standards

Where the coverage stops

No published standard establishes humanoid-robot battery-pack qualification requirements for energy density, pulse-power profiles, cycle life, bipedal vibration, installed-pack fall impact, human-proximity thermal runaway acceptance, or controlled-shutdown duration. An earlier piece in this section on drone/UAV pack qualification separated two different kinds of absence: a parameter governed by a standard whose text we could not retrieve publicly, and a parameter no standard governs at all. In humanoid robotics, nearly every performance and humanoid-specific parameter class falls into the second category.

As of 14 August 2026, no publicly identified working group across ISO/TC 299, IEEE RAS, A3/RIA, UL, or IEC TC 21 addresses humanoid-robot battery-pack qualification. China's Project 20261658-T-604 and the IEEE RAS Humanoid Robots Study Group take up humanoid safety and standards architecture in general terms. Neither publishes battery-pack test thresholds.

The classification question comes first

Before populating any DVP row, the program has to decide which IEC battery-safety standard it is building against, and neither standard resolves that for you.

IEC 62619 covers industrial applications. IEC 62133-2 covers portable applications. A factory-deployed humanoid supports a plausible industrial basis; a domestic or personal-care humanoid supports a plausible portable basis. This is a classification argument the program makes and the qualifying authority accepts. No clause in either standard supplies a humanoid-specific decision rule.

The downstream consequences are large. IEC 62619 contains a completed-system thermal-runaway propagation test (§7.3.3) and dedicated BMS protection tests (§§8.2.2–8.2.4) that IEC 62133-2 does not. IEC 62133-2 contains completed-battery vibration, mechanical shock, and free-fall tests (§§7.3.8.1, 7.3.8.2, 7.3.3) that IEC 62619 does not provide at the completed-system level. Pick one lane and you omit evidence the DVP needs from the other. Either way, the humanoid-specific parameters stay ungoverned.

Document the classification basis and the accepting authority before populating downstream rows. Where both standards contribute useful test evidence, propagation from IEC 62619 and mechanical environment from IEC 62133-2, cite both, provided the classification rationale supports it.

Requirements matrix by parameter class

Energy density and capacity

What the DVP needs: Gravimetric (Wh/kg) and volumetric (Wh/L) targets, minimum capacity, usable-energy reserve for controlled shutdown.

What governs it: Nothing. Neither IEC safety standard sets energy density, capacity, or reserve-energy thresholds. ISO 13482 §5.3.3 requires that power failure not create unacceptable risk, meaning dropped loads or uncontrolled movement, but sets no minimum shutdown duration or energy reserve (ISO 13482:2014, §5.3.3, per public reproduction; licensed text should control).

What's missing: The entire row. Mission duration, operating modes, controlled-shutdown energy budget, and mass/volume allocation come from the platform specification.

Power capability

What the DVP needs: Peak and sustained discharge C-rates by operating mode (walking, stair climbing, manipulation, balance recovery), pulse duration and repetition frequency, charge C-rate, regenerative-charge profile.

What governs it: IEC 62619 §8.2.3 verifies BMS response to 120% of maximum charging current (IEC 62619:2022, §8.2.3, per public reproduction). That tests protective response to overcurrent. It says nothing about operational discharge capability.

Published humanoid research provides platform-specific measurements rather than general requirements. TOCABI (48 V, 976.8 Wh) reported approximately 300 W idle, 600 W moderate activity, and slightly above 1,000 W during rapid whole-body actions, which works out to roughly 0.31C, 0.61C, and 1.02C at pack level (editorial: derived from reported power against pack energy, not measured as C-rate). HRP-2 stair climbing demanded approximately 45 A against an 18 Ah NiMH pack, a current-to-capacity ratio of roughly 2.5C on the same editorial basis. Both are configuration-specific results on research platforms with their own chemistries and pack sizes.

What's missing: No governing standard sets humanoid discharge C-rates, pulse durations, duty cycles, or regenerative-charge profiles. The DVP defines these from the platform's actuator power demands, pack architecture, and mission profile. What the literature does establish is that demand varies materially with locomotion mode, speed, stair height, and balance correction, which means constant-current testing will not represent humanoid service unless the program can demonstrate otherwise for its specific duty cycle.

Cycle life and calendar life

What the DVP needs: Cycle count at specified DOD, C-rate, temperature, and SOH retirement threshold. Calendar life at specified SOC and temperature.

What governs it: Nothing reviewed. Neither IEC safety standard, ISO 13482, nor any identified adjacent framework sets cycle-life or calendar-life acceptance criteria for humanoid packs.

What's missing: The entire row. DOD window, representative duty cycle, temperature range, SOH floor, and calendar-life storage conditions come from the platform's operational and commercial requirements.

Operating envelope

What the DVP needs: Operating and storage temperature range, humidity, altitude, condensation tolerance, recovery conditions.

What governs it: The IEC 62619 BMS overtemperature test (§8.2.4) verifies protection at 5°C above the manufacturer's declared maximum operating temperature. The manufacturer declares the limit; the standard does not prescribe it. ISO/TR 23482-1:2020 includes selectable robot-level methods for temperature/humidity fluctuation, combined environmental and vibration endurance, and low-pressure operation at normal, low, and high temperatures (Clause 9, per public TOC preview).

What's missing: No humanoid-specific operating temperature range, humidity limit, altitude ceiling, or condensation-recovery criterion. The DVP defines these from the intended deployment environment and flows them to the pack specification.

Mechanical — vibration, shock, fall/impact

What the DVP needs: Bipedal gait vibration spectrum at the pack mounting location. Installed-pack fall shock profile for a 1–2 m robot fall. Handling and transit shock.

Vibration — what governs it: IEC 62133-2 §7.3.8.1 prescribes a sinusoidal 7–200–7 Hz sweep, approximately 15 minutes per cycle, 12 cycles per axis, three axes. That is a portable-battery test profile. Published humanoid research reports walking fundamental frequencies near 1.73 Hz with harmonics to roughly 10 Hz, and sensor-mount disturbances around 46–55 Hz. These are angular dynamics or sensor-specific measurements, not translational acceleration PSDs at a battery attachment interface.

MIL-STD-810H Method 514.8, Procedure I provides a tailorable vibration framework that accepts measured or analytically derived environments. Its ground-vehicle categories do not establish a humanoid-torso spectrum. The program supplies the input PSD.

Shock and fall — what governs it: IEC 62619 §7.2.3 provides a mass-scaled handling drop (1,000 mm below 7 kg, down to 25 mm at 100 kg and above). IEC 62133-2 §7.3.3 drops a fully charged battery three times from 1.0 m. IEC 62133-2 §7.3.8.2 applies half-sine mechanical shocks (150 gn, 6 ms for the small-battery branch; check the mass/size classification in the licensed standard). None of these reproduces an installed pack carried into the floor by a falling articulated robot.

Published fall research reports trunk peak accelerations of 4.26–7.35 g for a 60 kg humanoid falling onto a mat under various control strategies, and 69–142 m/s² in a wall-support study. Both are geometry- and controller-specific results on non-representative surfaces, not standardized shock pulses at a pack interface.

MIL-STD-810H Method 516.8 provides tailorable shock procedures: Procedure I (Functional Shock) for demonstrating continued operation, Procedure IV (Transit Drop) for handling, Procedure V (Crash Hazard) for containment. The program defines the humanoid-representative pulse shape, axes, repetitions, mounting configuration, and post-test acceptance criteria.

What's missing: No public source provides a generally applicable bipedal pack-location vibration PSD, a representative installed-pack fall shock pulse, or humanoid-specific post-impact acceptance criteria. The DVP generates these from measured pack-interface dynamics, a validated multibody model, or another configuration-specific basis. Citing IEC 62133-2's sinusoidal sweep or a MIL-STD-810H vehicle category without a humanoid-derived input gives you a placeholder in the row, not qualification evidence.

Mechanical — crush/impact at cell and cell-block level

What the DVP needs: Completed-pack crush or structural-loading test representing forces during a robot fall or collision.

What governs it: IEC 62619 §7.2.2 provides a cell/cell-block impact test: 9.1 kg mass dropped 610 mm onto a 15.8 mm steel bar at 50% SOC. IEC 62133-2 §7.3.5 provides a cell crush test at 13.0 kN. Both are cell-level.

What's missing: Neither standard contains a completed-pack crush test. Do not cite cell-level crush or impact results as pack-level qualification.

Safety — thermal runaway propagation

What the DVP needs: Propagation test with trigger method, observation period, and acceptance criteria appropriate for a pack installed in a robot operating near people.

What governs it: IEC 62619 §7.3.3 is a completed-battery-system propagation test. The system is fully charged, stabilized at 25°C ± 5°C, one cell is driven into thermal runaway, and the system is observed for 8 hours. Acceptance criteria: no external fire from the battery system and no rupture of the system enclosure. Fire or rupture caused directly by the triggered cell is permitted. The 2022 edition allows laser initiation and alternative methods per informative Annex B (IEC 62619:2022, §7.3.3 and Annex B, per public reproduction; independently corroborated by JRC).

IEC 62133-2 contains no propagation test.

What's missing: IEC 62619's acceptance criteria were written for industrial battery systems in industrial installations. They do not address allowable external surface temperature during and after propagation, vent-gas composition or volume limits for occupied spaces, particulate or flame-jet exposure at human proximity distances, or time to safe condition. ISO 13482 §5.3.2 requires that stored energy not produce a hazard but sets no propagation-specific threshold.

The DVP has to define human-proximity acceptance criteria: heat flux, gas exposure, external surface temperature, safe standoff distance, evacuation time. For a pack intended to operate around people, this is the row to resolve first, because nothing in the borrowed test characterizes what happens outside the enclosure.

Safety — abuse tolerance

What the DVP needs: External short circuit, overcharge, forced discharge, thermal abuse at the pack level.

What governs it: Most IEC abuse tests are cell-level. IEC 62619 external short (§7.2.1), thermal abuse (§7.2.4), overcharge (§7.2.5), and forced discharge (§7.2.6) are cell or cell-block tests. IEC 62133-2 thermal abuse (§7.3.4), crush (§7.3.5), and forced discharge (§7.3.7) are cell-level. IEC 62133-2 external short (§7.3.2) and overcharge (§7.3.6) are completed-battery tests. IEC 62619 BMS protection tests (§§8.2.2–8.2.4) verify completed-system overvoltage, overcurrent, and overtemperature response.

What's missing: Do not cite cell-level abuse results as pack-level qualification. Where the program needs pack-level abuse evidence beyond what IEC 62619's BMS tests and IEC 62133-2's completed-battery tests provide, both the test conditions and the acceptance criteria are program-defined.

Environmental protection

What the DVP needs: IP rating, water/sweat/cleaning-fluid exposure, dust, salt fog.

What governs it: ISO 13482 §5.2.1 points to IEC 60529 for ingress protection where appropriate. ISO/TR 23482-1 includes selectable sealing and robustness methods (Clause 9, per public TOC preview).

What's missing: No universal humanoid IP class, sweat exposure, cleaning-fluid resistance, dust, or salt-fog requirement. Program-defined from the deployment environment.

EMC

What the DVP needs: Immunity and emissions levels for the battery pack.

What governs it: IEC 62619 Clause 9 requires the battery system to meet EMC requirements applicable to the end-use equipment or otherwise agreed for the application. ISO 13482 §5.8 requires safety functions to be maintained under relevant electromagnetic disturbances. ISO/TR 23482-1 includes EMC-related robot-level methods.

What's missing: The applicable levels flow from the end-product regulatory basis and the target jurisdiction, not from the battery standard. Cite the end-product EMC standard in the DVP row. IEC 62619 Clause 9 on its own does not give you a threshold.

Transport

What the DVP needs: Transport classification evidence for the battery pack.

What governs it: UN 38.3.

What's missing: Nothing for transport classification itself. A UN 38.3 pass does not close any installed-operation, fall, propagation, mission-power, or service-life row.

Gaps summary

Every row above except EMC, which is governed at the product level, contains at least one parameter the DVP has to define itself with documented rationale. Three rows have partial adjacent coverage: operating envelope through ISO/TR 23482-1's selectable methods, propagation through IEC 62619's industrial-context test, controlled shutdown through ISO 13482's safe-behavior requirement. None of the three supplies a humanoid-specific acceptance criterion. For the two humanoid-specific mechanical hazards, bipedal vibration and fall impact, the published research confirms the phenomena and gives platform-specific numbers, but nothing in it can be dropped into a DVP as a test condition without program-specific derivation.

The structural relationship is straightforward. ISO 13482 enumerates hazards for personal care robots and does not test batteries. The two IEC standards test batteries and were not written against those hazards. What falls between them is what the program owns, and for humanoid packs that is most of the matrix.

Two practical consequences. First, document the derivation behind every program-defined threshold at the time you set it, not afterward. Second, track standards-body activity for closure, because a standard arriving later does not validate what you chose in its absence. It creates reconciliation work: gap analysis between your values and the published thresholds, re-test wherever acceptance criteria differ, and a documentation pass to replace program rationale with clause citations. That cost scales with how thinly the original rationale was recorded.