Build an AIoT system which could deploy AI model automatically and continously from cloud to edge, collected and sent new valid data from edge to cloud. After cloud node continously receive new data and retrain model to deploy, users could get better service through the Cloud-edge life cycle.
- Thourgh the container lightweight and Kubernetes cloud orchestration advantages, users could easily scale the edge devices and no system restrictions.
Through automatic and continuous updating of AI model, users can continuously get better services. Besides, users can also change their services through the webpage, and choose and deploy the services which they need.
KubeEdge is based on Kubernetes. It provides the network between cloud and edge, and extends the scheduling for service from cloud to edge. This also called “the lightweight of Kubernetes”. In the architecture of AIoT, it divide the original cloud platform to cloud and edge which is closer to users. In compare with cloud, edge can provide more immediate service than cloud.
Continuously collecting a large amount of new data from edge node makes AI model accuracy higher. AI model is also used to filter invalid data by the identification result. In this way, it can reduce the computing resource when sending back and training on the cloud node.
About Yi-Cheng
r