SynapX Launches SYNData for Embodied AI

🔑 Enhanced Key Takeaways

• •SYNData is a component of SynapX's broader 'SYNTH' architecture, which also includes the SYNAction operational intelligence model and the SYNWorld physical world foundation model.
• •The system utilizes a proprietary 'Omni-modal Physical Data System' (OPDS) that introduces the concept of 'AI as a Sensor' to generate high-precision multimodal physical signals.
• •SynapX is a Beijing-based startup founded in January 2026 by Du Dalong, a former early employee at Horizon Robotics, and has secured nearly $50 million in seed funding from investors including Horizon Robotics, Xiaomi, and Shunwei Capital.

🛠️ Technical Deep Dive

• •OPDS (Omni-modal Physical Data System): Integrates proprietary hardware and AI algorithms to collect and generate multimodal physical signals (vision, force, touch).
• •REMA (Rhythmic End-to-End Manipulation Architecture): A multi-frequency, multi-scale framework within the SYNAction model consisting of three layers: System 2 (high-level reasoning/planning), System 1 (action strategy generation), and System 0 (low-level control/execution).
• •VFT-WFM (Vision–Force–Tactile World Foundation Model): A framework within the SYNWorld component that unifies vision, force, and touch into a single physical interaction model.

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline