UC2: Data-driven energy-efficient manufacturing

.
├── custom                # Java code for custom processor 
├── DAYTON.ttl            # dummy data for triple store
├── fuseki-configuration  # fuseki configuration file
├── input                 # dummpy input data for data flow test
├── NiFi_Flow.json        # exported NiFi flow file
├── nifi-jsonld-nar-2.0.0-M2.nar # created custom processor
├── README.md
└── simulate.py           # simulation code for workload genration

Introduction

This repository contains code and file related to UC2 data collection and workload generation.

• Robot navigates a manufacturing site, using Computer Vision tools (YOLO) to generate object detection results
• YOLO results (JSON files) are streamed into GLACIATION platform, more specifically, Apache NiFi data flow management tool running on the platform
• Apache NiFi has a HTTPListening component to take the input and store the results into DKG
• Example Workload: Construct images labeled with detected humans from stored results in DKG for manual evaluation

Requriements

• Apache NiFi 2.0.0-M2
• Custom processor needs to be in extensions folder

Apache NiFi data flow

• ListenHTTP is listening HTTP requests (e.g., sending JSON files via HTTP from robots)
• ExecuteGroovyScript decode base64 encoded images and store in another location specified in PutFile
• Semantification is a custom processor for semantifying the YOLO results according to the UC2 ontology below, and store the information into Apache Jena instance (DKG).

UC2 ontology for storing YOLO results

• Use case 2 ontology with some extentions following the upper ontology defined in T6.1