NSPIC-R007NP+ Embodied Intelligence Control Platform

联系二维码

Scan the code to enquire about the product

Product Description

The NSPIC-R007NP+ achieves deep fusion of "brain" cognition and "cerebellum" control on a single SoC, providing an end-to-end logic for embodied intelligence. It utilises a 14-core ARMv9 architecture and NIIC SYNC MANAGER for high-frequency synchronised perception. With a 2070 TFLOPs FP4 AI specification, it natively supports real-time decision-making for VLA/VLN models with up to 9 billion parameters. Combined with innovative multi-master EtherCAT technology and NMA acceleration, it ensures that AI-generated commands are translated into precise whole-body movements.

Product Specifications

Company Profile

Guoxun Micro (Suzhou) Technology Co., Ltd.

Guoxun Xinyi (Suzhou) Technology Co., Ltd. was established in October 2017 and specialises in full-stack control technology for embodied intelligence. The company focuses on the research, development and production of embodied intelligence operating systems and their underlying products, and is currently the only supplier in China to have realised an end-to-end closed-loop “perception-decision-execution” control platform on a single SoC. Leveraging its proprietary NECRO real-time operating system, NIIC EtherCAT Master 2.0 synchronous multi-master technology and NMA StarSpeed Matrix Accelerator, Guoxun Xinyi has established a deterministic computing and control architecture. In conjunction with the NIIC SYNC MANAGER multimodal synchronisation manager and the OCS2 full-body motion control/ROS2 hard real-time acceleration platform, the company has achieved deep integration of perception alignment, logical reasoning and physical execution, providing a hardware foundation for the real-time deployment of edge-side large models of varying parameter scales within the embodied intelligence domain.
Visitor Registration Call us WeChat

Leave Messages

Exhibitor Application
Visitor Registration Contacts Add WeChat

Reserve a Booth

Exhibitor Application

Medium/EN