The freight and transport logistics sectors are seeing ever greater use of technologies in warehouses, as well for stock-picking, consignment loading and delivery checks, among other applications, as the sectors work to automate a range of processes and improve the accuracy of those processes.

New capabilities were being created using existing systems that were integrated with new machine learning (ML), image recognition and AI solutions to deliver new insights, capabilities and controls, logistics risk management company Professional Risk and Asset Management group operations director Gerhard van Zyl said.

He was speaking during the Safety and Security in Transport and Logistics conference hosted by industry organisation the Chartered Institute of Logistics and Transport of South Africa on May 22, in Woodmead, Johannesburg.

Professional Risk and Asset Management has implemented a proof-of-concept solution for delivery surveillance.

"One of the things we are working on at the moment is to match imagery without human intervention from product loading onto the vehicle and then offloading," he told attendees.

"The solution uses closed-circuit television (CCTV) present in the warehouse and in trucks, along with ML to match what has been unloaded from the truck to what was loaded, and that the pallet that is being delivered is the correct one for that client."

The solution uses the drop sequence – the stops and deliveries along a truck's route – and matches that with the telematics on the vehicle to determine where a pallet is being delivered. The solution then uses the image of the pallet inside the truck to determine whether it is present in the truck, has not been tampered with and is the correct one that must be delivered to that client.

"Our goal is to automate the checking function in totality. CCTV provides the live monitoring of dispatch, and a combination of cameras and augmented reality (AR) wearable devices worn by pickers is used to ensure picking is correct and the entire load is guaranteed to be correct.

"The AR device worn by the pickers sends their picking orders to them and also monitors what is being picked and in what aisle. When the picker picks an incorrect item or in an incorrect aisle, the system will alert the picker and send an alert to the control room. This also provides proof that what is on the pallets loaded into trucks has been picked correctly.

"The livestream footage from CCTV and wearable cameras is monitored by the warehouse ecosystem management solution. We can integrate ecosystems to create outbound, inbound and en-route monitoring and control solutions," he said.

Further, Van Zyl highlighted that most video feeds and livestreams were monitored by ML- and AI-based systems. For example, Professional Risk and Asset Management had used vehicle licence plate recognition for access control in a Johannesburg-based warehouse.

"We are starting to rely more on AI to monitor these processes that need constant attention, such as video streams, so that we do not have to rely on a person for intervention. We use the AI systems to monitor videos and check picking, loading and delivery," he pointed out.

"Cameras are providing far larger functionality than only keeping eyes on the prize. Cameras can provide intelligence that we can use in a clever way."

Camera-enabled solutions were being used to map out warehouse flows to optimise warehouse management, and the integrated system could alert managers when stock was near expiration, or if there was old stock that should not have been picked and loaded, Van Zyl illustrated.

"We can use existing technology, but integrated and using AI to fulfil certain functionalities. The time-in-motion information means the layouts of warehouses can be reconfigured for optimal layout every two to three months using the information from the AR units while daily picking takes place," he said.

Additionally, companies also have footage of everything that has happened from receiving to delivery, if there are disputes.

"This is where the future of picking is. Certain companies, such as Amazon in the US, are already using such systems."

The client that Professional Risk and Asset Management ran the proof-of-concept for has developed the solution into a fully-fledged project for national rollout.

However, he warned that developing such a solution was an involved process that had to be tackled in a methodical way.

Companies have to ensure the basics are correct, from destacking and checking all goods from receiving, to ensuring that they have a good reverse logistics process in place, he advised.

"If you do not have adequate reverse logistics and you do not know that the inventory that you take in is correct, then a digital solution will not stop losses."