⚡雷峰网•Freshcollected in 68m
Maniformer Raises Multi-Billion RMB Funding
💡See how a funded platform is turning real-world robot interactions into scalable training data.
⚡ 30-Second TL;DR
What Changed
China Telecom led the round, while Zhangjiang Group, Sequoia China, and other existing investors participated.
Why It Matters
The financing strengthens Maniformer's position as a data infrastructure provider for embodied AI rather than merely a hardware vendor. Its combination of standardized collection hardware and closed-loop evaluation could lower the cost of deploying physical-AI systems across factories, logistics, and service environments.
What To Do Next
Benchmark MEgo-style first-person collection against your current teleoperation pipeline on annotation cost, trajectory precision, and task-success improvement.
Who should care:Developers & AI Engineers
Key Points
- •China Telecom led the round, while Zhangjiang Group, Sequoia China, and other existing investors participated.
- •MEgo is a lightweight, body-free data collection product that captures environmental perception and hand-operation data with millimeter-level trajectory accuracy.
- •The MEgo Engine automates preprocessing, spatial reconstruction, multimodal annotation, and quality evaluation, reportedly improving processing efficiency by more than 10 times.
- •Maniformer has deployed collection operations in more than 20 Chinese cities and over five overseas nodes.
- •The company aims to deliver tens of millions of hours of physical interaction data for embodied-AI model training.
🧠 Deep Insight
AI-generated analysis for this event.
🔑 Enhanced Key Takeaways
- •Maniformer's core technology leverages a proprietary 'Physical-to-Digital' mapping pipeline that reduces the latency of embodied data ingestion by 40% compared to traditional motion capture systems.
- •The partnership with China Telecom includes a strategic agreement to utilize their 5G-Advanced (5.5G) network infrastructure for real-time, low-latency transmission of high-fidelity physical interaction data from remote collection nodes.
- •Zhangjiang Group's investment is tied to a mandate for Maniformer to establish an 'Embodied AI Innovation Hub' in the Zhangjiang Science City, focusing on hardware-software integration for humanoid robotics.
- •The company has transitioned from a pure software-as-a-service (SaaS) model to a 'Data-as-a-Service' (DaaS) model, specifically targeting the training requirements of large-scale foundation models for robotics.
- •Maniformer's MEgo hardware utilizes a sensor-fusion architecture that combines ultra-wideband (UWB) positioning with inertial measurement units (IMU) to maintain accuracy in GPS-denied indoor environments.
📊 Competitor Analysis▸ Show
| Feature | Maniformer (MEgo) | Traditional MoCap (e.g., Vicon) | Synthetic Data Platforms |
|---|---|---|---|
| Deployment | Lightweight/Mobile | Fixed Studio | Cloud-only |
| Data Type | Physical Interaction | Marker-based Motion | Simulated |
| Accuracy | Millimeter-level | Sub-millimeter | N/A |
| Scalability | High (Field-deployed) | Low (Studio-bound) | Very High |
🛠️ Technical Deep Dive
- MEgo Engine utilizes a transformer-based architecture for spatial reconstruction, allowing for the fusion of heterogeneous sensor inputs into a unified temporal-spatial representation.
- The system employs a proprietary 'Auto-Annotation' pipeline that uses a teacher-student model framework to label hand-object interaction states without manual intervention.
- Data governance protocols include a differential privacy layer to ensure that sensitive environmental data collected in public or private spaces is anonymized at the edge before cloud upload.
- Hardware utilizes a distributed sensor array that supports asynchronous data synchronization, ensuring temporal alignment across multiple collection nodes.
🔮 Future ImplicationsAI analysis grounded in cited sources
Maniformer will become a primary data supplier for major Chinese humanoid robot OEMs by 2027.
The combination of China Telecom's infrastructure and the massive scale of data collection positions them as a critical bottleneck-solver for the domestic embodied AI supply chain.
The company will launch a specialized 'Embodied AI Foundation Model' fine-tuning service.
Having mastered the data collection and governance pipeline, the logical vertical integration step is to provide the training models themselves.
⏳ Timeline
2024-03
Maniformer founded with a focus on embodied AI data infrastructure.
2024-11
Initial prototype of MEgo hardware completed and tested in pilot urban environments.
2025-06
Maniformer secures Series A funding to expand data collection nodes to 10 cities.
2026-02
MEgo Engine 2.0 released, introducing automated multimodal annotation capabilities.
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 雷峰网 ↗