ARNEIS
The ARNEIS (short for Automated Recognizer, Network-Enabled, Item Sorter) project is the winner of the Popular Vote of the OpenCV Spatial AI Contest sponsored by Intel® and Microsoft Azure.
ARNEIS in a nutshell
ARNEIS aims at reproducing in scale a packaging machine for the Industry-4.0, thanks to a combination of:
- An OAK-D-Lite Spatial AI camera
- A lot of LEGO® Technic parts
- An intelligent combination of both existing and new Open Source software...
- ... deployed to a hybrid Kubernetes cluster made with resources
- partly on Azure cloud running on Intel® hardware
- partly on the edge using either a Raspberry Pi or other Embedded Linux devices.
ARNEIS Project documentation
Even though the machine is made of LEGO®, we strived to develop the project in a professional way as we are used in our daily job. With this in mind we have released at https://arneis.readthedocs.io/ the complete documentation of the ARNEIS project, including:
- functional specifications
- system and software architecture documents
- design documents with detailed instructions for reproducing the machine
- system integration and validation documents
However, as the old adage says (and we learned it the hard way)
here is a 3-minute video which we submitted to the OpenCV Spatial AI Contest final on 2022-04-04:
The source files for the software programs, the LEGO® MOC as well as the documentation site are maintained in the GitHub repository at https://github.com/B-AROL-O/ARNEIS.
Whenever the main branch of the git repository is updated, https://arneis.readthedocs.io/ will be updated accordingly.
The ARNEIS MOC
We have designed a MOC (My Own Creation) using the parts from LEGO® Set 42100 (Liebherr R 9800 Excavator).
The ARNEIS Software Architecture
The software architecture is based on microservices running on a hybrid Kubernetes cluster. Please refer to https://arneis.readthedocs.io for details.
The Computer Vision and Machine Learning
The OAK-D-Lite captures photos of the bottles flowing through the conveyor and runs a Neural Network trained on a custom dataset to be able to classify and recognize the type (Part Number) of the bottle. This information is fed in real-time to the Raspberry Pi which - based on the job order - decides whether to eject the bottle to the final packaging unit, or putting it back to the parts warehouse.
ARNEIS project roadmap and major milestones
The ARNEIS project roadmap is kept updated on GitHub.
AI Show Live - Episode 59, 2022-06-17
Seth Juarez interview to B-AROL-O Team starts at 13:11
B-AROL-O team interview, 2022-03-17
During OpenCV Weekly Webinar Episode 49 the B-AROL-O Team was asked to present the ARNEIS project and explain the vision, the motivations and the achievements after the first three months of development:
ARNEIS LEGO MOC History
A 5-min video with the initial design of the LEGO MOC is available on YouTube.
How to stay in touch
You may follow @baroloteam on Instagram or @baroloteam on Twitter X to get notified about the progress of the ARNEIS project.
Please report bugs and feature requests on https://github.com/B-AROL-O/ARNEIS/issues, or DM B-AROL-O Team on Twitter X about security issues or other non-public topics.
Copyright and license
Copyright (C) 2021-2024, B-AROL-O Team, all rights reserved.
Source code license
The source code contained in this repository and the executable distributions are licensed under the terms of the MIT license as detailed in the LICENSE file.
Documentation license
Please note that your contribution to the ARNEIS Documentation is licensed under a Creative Commons Attribution-Share Alike 4.0 License. see https://creativecommons.org/licenses/by-sa/4.0/