humanoidmetric.com
Independent humanoid-robot metrics · Xodexa
HMX v2.0.0
← Back to the Humanoid leaderboard

🇰🇷 KAIST Humanoid v0.5

KAIST Dynamic Robot Control & Design Lab · South Korea · unveiled 2025
research task-autonomous electric data confidence: medium
33.0
HMX Score
Rank #140 of 169 humanoid

Watch changes · Evidence history and cited research export (API access) · Plan a pilot

Score breakdown

AutonomyManipulationMobilityHardwareCommercialEcosystem
Autonomy & Intelligence · 25%92.0
Manipulation & Dexterity · 20%0.0
Mobility & Locomotion · 15%0.0
Hardware & Engineering · 15%56.9
Commercial Readiness & Deployment · 15%10.0
Ecosystem & Viability · 10%0.0

Each pillar is scored 0–100 from published specs, then weighted into the HMX. Compare it only with other humanoid robots — each category uses its own reference scales. See the full methodology →

Notable

KAIST Humanoid v0.5 was a precursor bipedal platform showcased in September 2025, demonstrating reinforcement learning-based locomotion. It was succeeded by v0.7 in 2026.

Specifications

Evidence is attached to each value. Robot-wide source links do not verify every field. A source about an older value cannot support the current specification.

Height
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Weight
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Payload
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Travel speed
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Runtime
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Battery
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Total DoF
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Hand DoF per hand
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Dexterous hands
No
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Actuation
electric
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Form factor
bipedal
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Onboard compute
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

AI model
Simulation-trained Reinforcement Learning (RL) locomotion policy
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Autonomy
task-autonomous
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Product status
research
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Indicative price
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Funding
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

Company valuation
Undisclosed
unverified · 0 evidence records

Declared shared origins are grouped. Distinct domains and unclassified origins do not establish independent evidence.

No field-specific evidence recorded. This value is unverified at field level.

KAIST Humanoid v0.5 in context: rank, rivals and specs

Compare with Leju Robot AELOS Series →

KAIST Humanoid v0.5 is a humanoid robot from KAIST Dynamic Robot Control & Design Lab in South Korea, and it is a research platform. It scores 33.0 on the Humanoid Metric, ranking 140th of 169.

On the leaderboard it sits between Leju Robot AELOS Series (HMX 33.2) and Honda E2-DR (HMX 32.9).

Against the 169 humanoid robots ranked, its clearest strength is autonomy & intelligence (92.0 against a median of 70.0) and its clearest weakness is commercial readiness & deployment (10.0 against 67.0).

On autonomy, it has demonstrated completing tasks on its own. No deployment is on record.

Its specifications carry a medium-confidence grade from 2 cited sources, as of 2026-08.

Task evidence

Task capability is not inferred from hardware specifications or general autonomy labels. Find robots for a task →

No task-specific evidence recorded. Capability remains unknown.

Frequently asked questions about the KAIST Humanoid v0.5

How much does the KAIST Humanoid v0.5 cost?

KAIST Dynamic Robot Control & Design Lab has not published a price for the KAIST Humanoid v0.5, and none is estimated here.

Is the KAIST Humanoid v0.5 autonomous?

The KAIST Humanoid v0.5 has demonstrated completing tasks on its own. The grade reflects demonstrations, not marketing claims.

How does the KAIST Humanoid v0.5 compare to the Leju Robot AELOS Series?

On the HMX, the KAIST Humanoid v0.5 scores 33.0 and the Leju Robot AELOS Series 33.2. The KAIST Humanoid v0.5 leads on autonomy & intelligence; the Leju Robot AELOS Series leads on manipulation & dexterity, hardware & engineering and commercial readiness & deployment.

Who makes the KAIST Humanoid v0.5?

The KAIST Humanoid v0.5 is made by KAIST Dynamic Robot Control & Design Lab, in South Korea.

Embed this score

HMX score: 33.0

Version history · 2 versions and 1 earlier update

Full history →

Nothing is deleted: every version of this record is kept, with what changed in each.

  • Version 335 · 15 Sep 2026, 04:28 UTC · score then 33.0
    No field changed; the score was recalculated.
    Changed by a direct correction or a scoring update, not a research proposal.
  • First recorded version · 6 Sep 2026, 05:24 UTC · score then 33.0
    The record as it stood when full copies began to be kept.
Data confidence: medium. Some figures for this robot rest on single or conflicting sources, or on manufacturer/aggregator material rather than independent reporting. Treat second-order specs as approximate. Fields marked “—” were genuinely undisclosed and left blank rather than estimated.