GuangheTong achieves 2.6 times acceleration of VLA edge inference

According to Guanghetong News, the Guanghetong AI Research Institute has recently made a key breakthrough in the field of embodied intelligence: based on the self-developed FiboVLA framework and edge-side inference optimization technology, the team has achieved an average 2.6 times faster inference on multiple mainstream VLA models, and completed the deployment of GR00T N1.5 on a high computing power control platform at the edge side. The related results have been validated through the LIBERO simulation benchmark dataset and desktop dual-arm real machine environment, providing engineering support for efficient operation of embodied intelligence models on robot edge side.