Australia Post is building a digital twin of its entire delivery network as part of continued efforts to use advanced analytics to detect and intervene on mail and parcel delivery problems.
The delivery network digital twin was revealed by Australia Post’s general manager of data science and strategy on a podcast earlier this month. Much of the work of the data science team that General Manager leads is well-documented.
The team, which was formed under finance but now sits in Post’s transformation and enablement function, is responsible for standing up several advanced analytics assets, including a data lake of sorts called Zoltar, named after the fortune-telling machine in the 1988 film Big.
More recently, it is responsible for Dexter, an “AI data bot” fed real-time data on mail movements that alert facility managers to potential issues.
The General Manager stated, “If your parcel is moving through the network and it’s due to be delivered today but we don’t see it get scanned onto a van by 6 am, the facility manager will start to receive emails from Dexter saying, ‘these are parcels we’re supposed to get out today, they’re somewhere in your network, go look for them’.”
However, it is the General Manager’s work on a multi-layered digital twin of the entire Post delivery network that is likely to be of substantial interest. Digital twins are digital copies of physical assets that are often used to plan and test future scenarios without impacting normal operations. The GM stated in the podcast that the big thing that the team is working on is a digital twin of the entire Australia Post network. That is huge.
The team is approaching it holistically. When people think of digital twins, they think of more scenario-based modelling but the team is thinking about it more like a grid, so three layers – an intervention layer, a forecasting layer and a simulation layer, and then interaction zones – an interaction with retailers, an interaction with ourselves in the network, and then an interaction with Australia Post customers.
AP’s data models have to fit within one of those grids, and then every model they develop now has to be part of what they are calling the digital twin ecosystem. “It has to have a life that contributes to that ecosystem, and then over time we will have eventually built a digital twin of the network,” the GM said.
The delivery network digital twin appears to be the second digital twin project at Australia Post. Having hinted at producing a virtual reality tool that could help posties complete difficult or unfamiliar delivery rounds back in 2018, a proof-of-concept emerged in October last year.
A professional services organisation said on its website that ‘Parcelbot’, as the proof-of-concept is called, also counted as a digital twin environment. It said it worked “in partnership with Australia Post” and used a mix of virtual reality technology, a virtual assistant AI technology an American multinational technology company and the Unreal gaming engine to create the tool.
The PoC: create a digital twin environment for posties to capture and surface important information along their delivery routes, including customer preferences like safe to leave a parcel unattended, locked gate, and protective dog.
The unlisted video accompanying the PoC shows how a postie can ‘look’ at an address and immediately see an overlay of information, such as recorded notes about the residents, how many parcels they receive, and how often missed delivery cards need to be left at that address.