For orchestrating WebAssembly-based Edge AI
sensing pipeline.
WEdge
Agent
New open-source Edge AI Vision sensor project for trillion-sensor network and sensor fusion, lowering the entry hurdle for developers ultimately.
AI-capable tiny/nano edge devices will exist everywhere in the future IoT environment. Once implemented, they remain there for a long period of time, while diverse sensor hardware, software, and networks work together to create/deploy/update novel solutions such as Smart Society, City, Factory, and other Smart-X services.
Challenges to solve
Tiny/nano edge devices need to be low-power and cost-effective, but with constrained resources
Agile Development for Resource Constrained IoT Devices
Embedded programming is difficult, generally in C, and generally the code is not portable
Safely running workloads on RTOS
Tiny IoT devices cannot run Linux, because their CPUs generally don’t have MMU
Avoid lock-in to specific IoT Service
Dependence on specific technologies would have created serious business continuity problems
WEdge
Technology overview
1
Easy to
develop
Integration of sensors, AI, CV, and processing at Far-edge/edge/cloud in a zero-touch/plug-n-play manner:
Take advantage of WebAssembly for resource-constrained devices
Sensing pipeline of simple function wasm modules, to be low-code/no-code
Easy to
deploy
2
Orchestrator: deployment, module life-cycle management
Webassembly compilation service: dynamic and automatic AoT optimization
Sensing pipeline deployment across different vertical locations (far edge devices, MEC, etc)
3
Easy to
enhance
Zero-touch provisioning and automatic edge service discovery enabled by integrating with telecom infrastructure intelligence.
Dynamic QoS control, QoS on Demand, Selecting best MEC location
Next events
Get to know our project at the following events
Embedded Vision Summit
Santa Clara, USA
Embedded Vision Summit
Santa Clara, USA
Embedded Vision Summit
Santa Clara, USA
Embedded Vision Summit
Santa Clara, USA