# embodied intelligence
Latest news and articles about embodied intelligence
Total: 9 articles found

As Cloud Giants Duel, China’s Mianbi Pushes a 9‑Billion‑Parameter Multimodal ‘Brain’ for Edge Devices
Mianbi Intelligence has launched MiniCPM‑o 4.5, a 9‑billion‑parameter multimodal model that can ingest continuous audio, video and text while producing simultaneous outputs, and introduced a Jetson‑based developer board called Pinea Pi. The company positions the stack as an early example of on‑device, embodied AI aimed at robotics, automotive and personal devices, arguing that hybrid cloud‑edge deployments will better meet latency, privacy and stability needs than cloud‑only approaches.

Chinese Startup Says the Dawn of Truly Natural Human–Machine Interaction Is Two to Three Years Away
Mianbi Intelligence’s founders say multi‑modal, embodied human–machine interaction is entering a decisive phase: rapid improvements in compact, dense models could bring meaningful advances within two to three years. The transition will be incremental, hinging on better on‑device models, cloud–edge integration, and solutions to privacy and power constraints.

China’s Fourier Says Brain‑Controlled Exoskeletons for Stroke Rehab Could Reach Clinics in 1–2 Years
Fourier, a Chinese robotics firm, says it will integrate brain‑computer interfaces with exoskeletons to enable robots to detect patients’ movement intentions and provide timed physical assistance, aiming for clinical rollout within one to two years. The move rests on more portable BCI hardware, AI advances in signal interpretation, and a staged market strategy that targets hospitals first, then care homes and households.

From Thought to Motion: Why China’s Robotics Industry Is Betting on Brain–Computer Interfaces for Rehabilitation and Elder Care
Chinese robotics firms are prioritizing rehabilitation and eldercare as the first real markets for embodied intelligence, pairing non‑invasive brain–computer interfaces with exoskeletons to restore patient agency and create quantifiable training data. Industry players are addressing technical, data and integration challenges through consortiums and clinical partnerships, aiming to convert early prototypes into scalable clinical products over the next three to five years.

Li Auto’s Big Bet: From Electric SUVs to Humanoid Robots and the Push for ‘Embodied Intelligence’
Li Xiang told Li Auto staff that the firm will evolve from an electric‑vehicle maker into an "embodied intelligence" company, pursuing base models, inference chips, an OS and humanoid robots. The strategic shift reflects investor pressure for higher‑margin AI products, recruitment of robotics talent, and significant technical and regulatory hurdles ahead.

China’s Magic Atom Leaps from Lab to Limelight — But Commercial Robot Reality Remains Costly and Long-Dated
Magic Atom, a young Chinese humanoid‑robot start‑up, was named an official robotics partner for China’s 2026 Spring Festival Gala, a high‑visibility sign that humanoid robotics are moving toward commercialisation. Its co‑founder warned that embodied intelligence remains a long‑cycle, capital‑intensive field: hardware costs, scarce effective real‑world data, and the difficulty of passing training costs to customers are the industry’s core challenges.

Suzhou Targets Semiconductors, Embodied AI Robots and Low‑Altitude Economy in New Five‑Year Blueprint
Suzhou’s recommendations for its 15th Five‑Year Plan mobilise the city to cultivate semiconductors, embodied intelligence robots, advanced machine tools and a low‑altitude economy, aiming to upgrade existing industrial clusters and create new growth poles. The move reflects a local strategy to deepen technological capabilities, scale new industries through demonstrations and integrate more closely with the Yangtze River Delta’s aerospace ambitions.

Chinese Startup Says It Trained Robots by 'Dreaming' — and Grew to Hundreds of Millions in Revenue
KuaWei Intelligence open-sourced EmbodiChain, a generative simulation pipeline that it says can train robot models entirely on synthetic data and achieve zero-shot transfer to the real world. The company reported annual revenue in the hundreds of millions of yuan for 2025 and expects three- to fourfold growth in 2026, pitching its approach as a scalable, ROI-focused alternative to costly real-world data collection.

As Models Mature, Chinese AI Firms Put Data — Not Parameters — at the Centre of Industrialisation
At the 2026 AIGC Developers Conference, Hangzhou Cangjie Intelligent’s CEO argued that industrial AI’s bottleneck is data, not models. Companies that build scalable, task‑structured, reusable data systems will gain the competitive moats required to deploy AI in manufacturing, robotics and embodied systems.