Korea launches 50.4 billion won humanoid programme with KIST and three LG affiliates, putting solid-state batteries in robots
Korea's science ministry on May 18 started a five-year, 50.4 billion won programme to build a national AI humanoid platform led by KIST, with LG Electronics, LG AI Research and LG Energy Solution among the partners. The plan includes a vision, touch, language and action model, an all-solid-state battery for humanoids and a field test of at least 20 robots in care settings.
South Korea's Ministry of Science and ICT on May 18, 2026 launched a five-year programme to develop what it calls a representative Korean AI humanoid platform, holding a kick-off meeting at the Korea Institute of Science and Technology (KIST) in Seoul, according to the ministry's press release.
The programme, formally titled the public-private AI humanoid source technology advancement project, runs from 2026 to 2030 with a total budget of 50.4 billion won, of which 35.4 billion won is government money and 15 billion won comes from private participants. The ministry made it a core project of K-Moonshot, a national effort to use AI to accelerate research across 12 missions that also include new drugs, nuclear energy and AI scientists.
- AI humanoid source technology (K-Moonshot), total
- Science and ICT
- 50.4
- 2026-2030
- Of which government
- Science and ICT
- 35.4
- 2026-2030
- Of which private
- Science and ICT
- 15
- 2026-2030
- Intelligent robot dissemination programme
- Trade, Industry and Energy
- 76.7
- 2026
- Humanoid parts demonstration (supplementary budget)
- Trade, Industry and Energy
- 5
- 2026
- M.AX dedicated fund (target size)
- Trade, Industry and Energy
- 500
- target
Figures as disclosed by the science ministry (May 18, 2026) and the trade ministry (May 29, 2026); 1 billion won = 10 억 won. Not estimates.
As of Oct 1, 2026
Who is building what
KIST leads the consortium. Industry partners are LG Electronics, LG AI Research, LG Energy Solution, the industrial-robot maker Robostar and the wearable-robot company WIRobotics. Seoul National University, KAIST, Korea University and Kyung Hee University supply academic teams, and Hallym University Sacred Heart Hospital will host field trials.
The hardware work starts from KAPEX, a humanoid that KIST developed in house. On that basis LG Electronics will develop a next-generation humanoid model designed for mass production, while WIRobotics will advance a mobile humanoid platform for use in public environments, the ministry said.
The technology
The intelligence work aims at a next-generation AI model that integrates vision, touch, language and action, so that a humanoid can perceive its surroundings, understand contact and force, and plan and execute complex tasks by itself. The programme outline calls this a VHLA model, adding touch, or haptics, to the vision-language-action models that dominate current robot AI. For 2026 to 2028 the outline lists VHLA-based motion and intelligence, a multimodal model for task execution and spatial awareness, and a whole-body humanoid hardware platform.
LG Energy Solution plans to apply what the ministry described as a world-first high-safety all-solid-state battery to the robot platform, with the aim of lowering fire risk, supporting long working periods and getting ahead on international safety standards. Battery safety is a practical issue for humanoids, which carry large packs close to people.
From 2029 to 2030 the consortium will deploy at least 20 humanoids in real settings such as hospitals and residential facilities for older people, with a target of continuous operation for more than one month, to test long, multi-step tasks that help with daily living and public services, the outline said.
Why the LG line-up matters
The programme puts three LG affiliates in one government project covering the robot body, the model and the battery. That gives LG a state-backed path to a mass-production humanoid alongside its consumer and commercial robot businesses, and gives LG Energy Solution a reference customer for solid-state cells outside cars. Kim Sung-soo, head of R&D policy at the ministry, said the project bundles AI, humanoids, batteries, mass production and field testing to build a national flagship platform.
Competitive context
The budget is small by global standards. Private humanoid developers in the United States and China have raised individual rounds many times larger, and Chinese makers are already shipping units at scale. Korea's bet is that a package combining a domestic battery champion, an electronics manufacturer and clinical test sites can produce a platform that is safer and better suited to care work, a market Korea's ageing population will expand.
Risks and open questions
The ministry did not disclose how the 50.4 billion won is split between partners, what specifications the mass-production model must meet, or when LG Electronics expects to sell it. A field deployment of 20 robots in 2029 and 2030 is late relative to rivals' commercial timelines, and the claim of a world-first solid-state battery for humanoids has not yet been backed by published cell data.
ROBOTNESS analysis
The programme is less a humanoid race entry than a state-funded integration test for LG's robot, model and battery businesses, and its value will be measured by what LG commercialises, not by the KAPEX prototype.
The evidence is in the structure. Three LG affiliates cover the body, the AI model and the energy system, the hardware role assigned to LG Electronics is explicitly a mass-production model, and the field tests run in care settings where a large electronics brand has distribution.
The strongest counter-argument is that a 50.4 billion won budget over five years cannot shape a market where private rounds dwarf it, and that state programmes with many academic partners tend to optimise for papers and demonstrations rather than products.
Bull case. LG uses the programme to de-risk a care and service humanoid with a safer solid-state pack, and LG Energy Solution turns robots into a second market for its next-generation cells. The VHLA model gives Korea a distinctive touch-aware capability that general vision-language models lack.
Bear case. Timelines slip, the 20-robot trial in 2029 arrives after Chinese and US rivals have set prices and standards, and the platform remains a research reference. The money is too thin to matter commercially.
Signals to watch.
- Any LG Electronics humanoid product announcement or development milestone before the 2027 to 2028 window.
- LG Energy Solution disclosures on solid-state cell specifications for robots.
- The 2027 science ministry budget line for K-Moonshot missions, due with the government budget proposal in the autumn of 2026.
- KIST, based on its KAPEX humanoid
- 50.4 billion won (35.4 billion government, 15 billion private)
- 2026 to 2030
- All-solid-state battery from LG Energy Solution
- VHLA (vision, haptics, language, action)
- May 18, 2026, kick-off at KIST
- Ministry of Science and ICT
- At least 20 humanoids, over one month, 2029 to 2030
- LG Electronics, LG AI Research, LG Energy Solution, Robostar, WIRobotics
- SNU, KAIST, Korea Univ., Kyung Hee Univ., Hallym Univ. Sacred Heart Hospital
Why it matters
This is the first Korean humanoid programme in which a single conglomerate's affiliates hold the body, model and battery roles under one state contract. That concentration matters more than the budget, because it lines up LG to commercialise whatever works.
The programme also introduces touch as a first-class input in a national robot model. Most current robot foundation models are vision-language-action models, and adding haptics targets the contact-rich work, such as helping people in care settings, where vision alone fails.
Rival analysis
In Korea, Samsung Electronics holds a controlling stake in Rainbow Robotics, and Hyundai Motor Group owns Boston Dynamics. This programme gives LG a comparable public anchor for its own humanoid effort.
Globally, the budget is modest next to private rounds, so the relevant comparison is not money but positioning. Care and residential support is a niche where safety certification and battery behaviour weigh more than raw speed, which suits a company with a battery affiliate.
Valuation context
The 50.4 billion won is not material to LG Electronics or LG Energy Solution financially. Its value is option value, a funded route to a mass-production humanoid and a robot battery reference.
For smaller participants such as Robostar and WIRobotics, a role in a flagship national programme is more significant and may support future fundraising or contracts, though the ministry did not disclose each partner's share.
Supply-chain implications
LG Energy Solution's role is the supply-chain story. A humanoid-grade all-solid-state pack would be one of the first non-automotive uses of the technology and could set a safety reference for international standards, which the ministry says is an aim.
On the hardware side, KAPEX gives the consortium an existing actuator and mechanism base, while LG Electronics brings appliance-scale manufacturing. Whether LG uses in-house motors or external suppliers is not disclosed.
Signals to watch
First, an LG Electronics humanoid unveiling or roadmap that references the programme. Second, solid-state cell specifications from LG Energy Solution for robots.
Third, results from the VHLA model in the 2026 to 2028 phase, which would show whether touch-aware control is working. Fourth, the 2029 start of the 20-robot trial at Hallym University Sacred Heart Hospital and other sites.
Analyst view
Thesis: the programme's main output will be LG commercial products rather than a national open platform. Confidence: medium. Reasons: the consortium design and the explicit mass-production role favour LG, but the ministry has not disclosed IP ownership or licensing terms.
The bear risk is timing. A care trial in 2029 and 2030 is slow next to rivals, and if LG does not move faster on its own, the programme will produce a reference design rather than a market entrant.
Questions you should be asking
Who owns the IP for the KAPEX-derived mass-production model, and can other Korean firms license it? What energy density and cycle life will the solid-state robot battery achieve?
Will the VHLA model be released to Korean industry, or kept within the consortium?