Swedish telecoms group Telia Company has developed a new machine vision solution for inventory management in warehouses which runs on a hybrid (‘virtual’; semi-private) private 5G network. The project, part of its work with Finnish logistics group Transval, automatically tracks and reports on the movement of inventory, removing the need for manual inventory counts and reconciliations, said Telia. AWS assisted on the software and edge development.

Transval, the biggest intralogistics house in the Nordics, part of Posti Group, commissioned Telia to devise a solution to automate inventory-checking in and out of a warehouse it managed on behalf of a customer. The proof-of-concept saw high-definition cameras installed and connected on a “virtual private network” to Telia’s data center in Helsinki. Amazon’s SageMaker cloud ML platform was used to build, train and deploy a model to recognize and track pallets.

Specifically, the solution was deployed on a ‘production buffer’ process at the warehouse, where adjustments are calculated for variations in production. The ML model calculates the number of pallets in a given area, reports how much free space is left in each stack and tracks the movement of pallets from one stack to another, revealing an accurate picture of inventory levels in real time, said Telia in a statement.

The old manual process was prone to human error, “characterized by counts and recounts”, it said. It explained: “Telia created an API that Transval used to develop an application that has been installed on its mobile devices, enabling employees to easily receive alerts from the ML model while moving around the warehouse. By ensuring that inventory counts match… the ML model prevents shortages that could otherwise bring the production line to a halt.”

Telia used AWS’s ISV Accelerate ecosystem sales scheme, it noted, to collaborate on the final solution. It also used AWS’s enterprise-edge (AWS Outposts) and local- and regional-edge (Local Zones and Availability Zones) platforms to bring compute and analytics resources closer to the customer. The understanding is it did not utilise its network-edge (MEC; Wavelength) service, and that the ‘virtual private network’ hinged mainly on public infrastrcture.

A statement said: “This use case could not have been developed if Telia had not invested in its 5G and modernized 4G networks, as well as in upskilling 10 percent of employees on AWS and cloud tech.” Telia said it will “standardize the development” of the solution to inform other bespoke Industry 4.0 projects (“enable other customers to achieve improved levels of automation and production optimization”) – if not to commercialise the same Transval solution.

Riku Koskimaa, senior manager for business development at Transval, said: “The solution developed… has increased our ability to access real-time information about buffer inventory levels in our warehouse. As a result, it has become much easier to ensure that our delivery buffer levels are as agreed with our customer and according to our KPI targets. By visually tracking buffer levels and providing real-time updates, we ensure customers have the items they need in their buffers at all times.”

Rainer Deutschmann, group chief operating officer at Telia, said: “Our collaboration with AWS to deploy our private networks offering combined with the cloud delivers simplicity, improved operational efficiency and the opportunity to quickly develop new initiatives to meet market needs. It’s about achieving scale and prioritizing security while reducing complexity, to the benefit of our customers and, ultimately, end users in our markets.”

Fabio Cerone, managing director for the telco industry at AWS in the EMEA region, said: “Real-time and high-fidelity experiences like warehouse logistics and supply chain require security and low latency. Leveraging AWS edge services and machine learning services, Telia was able to build, deploy, and manage a ML model and tailored private network to support Transval’s real-time needs.”