使用 AWS IoT Greengrass 设计本地生成式 AI 推理.pdf

1、 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.D E V 3 1 6Kohei“Max”MATSUSHITADesigning local Generative AI inference with AWS IoT GreengrassHe/H

2、imTech.Evangelist|An AWS HeroSoracom,Inc.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Real-world environments Physical AILocal vs Cloud Trade-of

3、fs and DesignDEMO Teleoperation Latency comparisonOperating and Updating Local AI model w/AWS IoT GreengrassDemo Model Update at the LocalDesigning Sustainable Physical AIAgenda 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Physical AI enables machines to sense,decide,and physic

4、ally interact with real-world environments.Local(Edge)AI and Cloud AI:Implementation layers where inference and updates happen.Physical AIPhysical AIThe concept of empowering real-world with AIEdge AICloud AIThe architecture and placement of AI inferenceInference where data isproducedInference where

5、 data iscollected 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Physical AI enables machines to sense,decide,and physically interact with real-world environments.Local(Edge)AI and Cloud AI:Implementation layers where inference and updates happen.Physical AIPhysical AIThe concept

6、 of empowering real-world with AIEdge AICloud AIThe architecture and placement of AI inferenceInference where data isproducedInference where data iscollected 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.ResponsivenessAutonomyCollaborationAI can act quickly when it sees or sense