bmpotter edited this page Oct 10, 2018 · 30 revisions

Welcome to the open-horizon examples wiki.

This wiki is a central resource for getting started with IBM's edge solution. You'll find more information at the links below about Edge compute, registration on the Edge system, and custom service development.

Getting Started

Horizon is an autonomous (zero ops) peer-to-peer (coordinated) edge compute platform that enables delivery of software to the edge in a general purpose, open, secure, multi-workload, multi-tenant fashion.

The above guides are based on this example code:

Deep Learning and AI-related Resources

  • NVIDIA Jetson TX1/2
    • cogwerx-jetson-tx2: open-horizon's Jetson TX2 github repository with code examples, Docker container builds for Yolo, Caffe, OpenCV, NVIDIA DIGITS, TensorRT
    • face-classification-tx2: a Tensorflow / Keras based CNN model, from oarriaga's open source codebase, packaged in Docker for TX2
    • self-jetsontx2: IBM Watson-Intu's open-source AI Middleware project, "Self". A Cognitive Middleware solution for tying together edge and cloud-based services. Read up on Intu and Embodiment here.

Microservices & Workloads by Architecture

Workloads typically process data and/or stream data to off-device endpoints such as IBM Watson IoT Platform.
Workloads leverage microservices at the edge, and are defined to run in Docker container groups.
Microservices provide access to hardware (cpu, network, camera, microphone) and raw data on the device.
Patterns define deployment of Workloads and Microservices for one or more architectures.

Microservice Name Workload Name ARMHF (RPi2/3) ARM64 / AARCH64 (TX2, C2) AMD64 (x86_64)
GPS GPS X X X Location (GPS device, IP-derived, or manual input), streamed to WIoTP via workload
CPU Percent CPU2WIoTP X X X CPU Load Stats, streamed to WIoTP
Network Netspeed2WIoTP X X X Network ping/speed test stats, to WIoTP
Weatherstation PWS2WIoTP X X Personal Weather Station data stream to Weather Underground and WIoTP
Pi3-Streamer PI3Streamer2WIoTP (Pi3 Only) MJPG Pi3 Camera LAN stream

Events

NVIDIA-IBM Edge Webinar

IBM's edge solution was presented as part of NVIDIA's Developer webinar series, demonstrating how to develop and run Edge compute services, including Deep Learning workloads.

Troubleshooting

You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.
Press h to open a hovercard with more details.