, AI Computer Vision: Powering Manufacturing

AI Computer Vision: Powering Manufacturing

Author: Philippe Glynn

Computer vision is a field of AI focusing on the ability of computers to interpret the real world through images, video, and other sources. With use cases across many different verticals, including transportation, automotive, healthcare, retail, education, and manufacturing to name but a few, computer vision is driving efficiency and streamlining processes.

Critical for computer vision use cases to succeed are the underlying infrastructure and IT environments in which they operate. To achieve true real-time processing, high reliability, and near-zero latencies are key characteristics. Like the human eyes and brain, the cameras capturing the data and the applications processing that data need to be collocated ensuring split-second decision-making.

Aspire Technology has developed a reference blueprint to drive computer vision use cases within the manufacturing industry. Incorporating the latest 5G cellular technology coupled with mobile edge computing to bring any computer vision use case to life. Partnering with industry-leading hardware and software companies, the reference blueprint is a lightweight solution that can be deployed seamlessly within any indoor and/or outdoor environment. The deployment targets existing IT environments thus leveraging on already established security principles.

Cellular 5G is the fifth-generation mobile network. 5G promises to deliver ultra-fast broadband (Gbps), ultra-reliable low latency communications (uRLLC), and provide the infrastructure to connect millions of IoT devices per meter squared. The 5G lean air interface has also been designed with sustainability in mind with the energy per bit transmitted is greatly improved over its predecessors.

Multi-Access Edge Computing (MEC) or Mobile Edge Compute is an ETSI defined framework and architecture, that enables the placement of cloud computing resources at the network edge. Placement of cloud resources at the network edge creates an IT environment where all end-to-end network functions are collocated, ensuring minimization of the data path between client and server, and resulting in near-zero latencies. The placement of Edge resources within a trusted IT environment also enhances security and greatly improves reliability as the end-to-end network is operating with known transport capabilities across a trusted network.

The full turnkey solution incorporates both SW and HW from Industry leaders enabling seamless end-to-end deployment. The underlying 5G network comprises Benetel’s RAN550 indoor n78 O-RU, IS Wireless vRAN (O-DU and O-CU) underpinned by VMware and Druid Software Systems modular 5G SA (Stand Alone) Core Network deployed using Intel® Smart Edge Open Developer Kit.

Leveraging Intel’s Edge Software Provisioner (ESP) enables automation of the installation of a complete operating system and software stack on bare-metal or virtual machines using a “Just-in-Time” provisioning process. Used to quickly boot up a bare-metal server with all predefined operating systems and software stack.

Intel® Smart Edge Open (SEO) is an edge computing software toolkit for building platforms optimized for the edge. Platforms created with Intel® Smart Edge Open can host a wide range of services, from network functions such as 5G RAN and 5G core to AI, media processing, and security workloads. Used as a starting point to create an edge location platform with K8s and cloud-native components, simplifying the deployment of our complex network architecture.

Aspire Technology provides all wrap around services enabling rapid deployment of the reference blueprint. This includes full radio planning, design and O-RU placement, maximizing coverage, capacity, mobility, and reliability. Managed services are executed remotely ensuring optimal execution 24/7/365.

Using private 5G spectrum, which is becoming more and more readily available to use across an ever-growing list of jurisdictions, a full turnkey solution can be delivered without the complexity and commercialization of spectral assets. The full solution can be deployed seamlessly and swiftly, once the radio planning and design is complete and the radio assets deployed according to the locations obtained in the planning. Although currently deployed as an indoor solution, outdoor deployments, or a combination of indoor and outdoor can also be deployed using Benetel’s outdoor O-RU RAN650 solution, covering several different band options for global deployments.

Our use case has been deployed within Aspire’s Open Networks Lab in Dublin and uses an IoT capable camera to continuously monitor infra-red sensors emulating a production line environment. The video feed from the camera is sent over the 5G air interface, vRAN and core network user plane function (UPF). Finally, the video feed is routed to the AI powered anomaly detection application over the N6 interface, which rapidly and accurately determines whether the infra-red sensor is substandard or faulty and for the purpose of this use case labels the sensors visually as such. Given the video feed from the camera will not consume the full resources of the underlying 5G network, other users accessing the 5G capabilities will be routed towards the internet or can be directed towards another private data network (DN).

Use cases can be tailored, delivered, and deployed to address the needs of any manufacturing environment and can also be extended across other verticals. Examples of such include computer vision to detect objects for self-driving vehicles, autonomous medical scan detections through to enabling cashless retail environments. The underlying infrastructure does not need to change and thus the benefits of the seamless and lightweight deployments are maintained.

For further information or to arrange a virtual or physical demonstration please contact us.


Link to original blog post AI Computer Vision: Powering Manufacturing