Urban Logistics: Turning the 'last mile' on its head

Package deliverers, as the name suggests, deliver packages. But before they can deliver, they must first sort the packages and load them into delivery vehicles. This is no small task: "Before a package delivery driver starts his route, he has usually literally carried at least a ton of packages for over two hours in the morning," explains Daniel Jarr, the head of Innovation at Dematic. It is a heavy physical burden and it takes a comparatively long time to get it done. It is quite inefficient. Isn't it possible to automate something, here?

This was precisely the question a customer posed to Dematic some time ago. As an expert in integrated supply chain automation technology and automation software, the company already has solutions for numerous warehouse processes in its portfolio. But some parts are not so easy to automate and this includes the final preparation step for the package delivery staff. What is pretty amazing is that package centers invest large sums in modern sorting systems that are also very efficient in delivering packages to the departure dock for the respective drivers. Yet no one had come up with a way of then packing the collected freight on a shelf in as little space and time as possible.

The statement highlights the situation around the "last mile", where many work steps are still characterized by tough physical labor. Deliverers maneuver their vehicles as close to the front door as possible, personally handing over the delivery and they are also busy in the mornings and evenings lifting packages into their vehicles or returning undeliverable goods. After an initial query to Dematic, it quickly became clear to project leads that this is a fundamental problem impacting numerous suppliers from various industries - from the package delivery service in retail trade as well as the supplier of spare parts. Jarr explains that he began doing field research by talking to delivery employees on the street: "It soon became clear that the two hours spent loading in the morning was very common."

The solution Dematic has developed to address the problem, “PackMyRide”, involves an automated system that can actually take packages off the conveyor belt and then uses algorithms to tightly fill a shelf. Dematic drew partially on various in-house developments. "These are, after all, core capabilities at Dematic," notes Jarr matter-of-factly. "On the one hand, you have 3D packing algorithms, which can also calculate which goods can be stacked on others, and then there's the automated system itself. However, in this case, the challenge was that these warehouse systems usually work with a rather clearly defined product range. They know already which products will come before their robot fingers. In contrast, when it comes to package delivery personnel, the product range can quickly become incredibly broad and it is usually quite random. "You can only determine the data to a limited extent in advance," explains Jarr. Weight is collected by packaging distribution centers as standard; yet, other data, such as size or stability, is not collected, or just to an insufficient degree and too late in the process. "So we had to teach the system to re-measure itself."

Next, the system had to learn how it grips and positions different-sized products. The question of the best gripping technology for robotic arms is one that countless engineers and companies have been looking into. Not surprisingly, the answer is that it depends on what exactly needs to be gripped. "However, in this case, it was also a question of how best to sort the package," notes Jarr, adding, "Imagine a large robot gripper arm that is supposed to sort a 10-centimeter small package into a standard bookshelf in the open gap next to a larger package. The gripper would be much too large for this, which is why with 'PackMyRide', Dematic decided on a concept based on very thin fingers that can push packages into narrow openings using a ‘flat hand’ type of movement."

All this was created on the proverbial "blank sheet of paper process". As Jarr puts it, "In the beginning, we had something like 50 to 100 ideas and then we split up into teams that competed against each other for a while," he says. In the end, they came up with about six solutions whose strengths were weighed against each other, leaving the one now known as ‘PackMyRide’ standing alone.

Dematic next tested the pilot system with its project cooperation partner, international package delivery service provider DPD. PackMyRide needed to prove it could handle a couple of challenges all at once: physically relieving the drivers and saving time, which, in turn, can be invested in delivery. "We have managed to load a good 75 percent of DPD's entire package spectrum fully automatically, which is an extremely good result for a first pilot system," states Jarr proudly. However, Dematic's project managers noticed that they obviously struck a nerve. The package shipping industry is projecting double-digit growth in the coming years and it also is anticipating that it cannot cover the additional demand for employees, given the current job market.

"The industry will inevitably have to look at other aspects, such as how to optimize a driver’s delivery times and what daily driver activities could be handled differently," suggests Jarr. What’s even more surprising for the Dematic project team is that there’s nothing rivalling what the PackMyRide pilot now makes possible. Yes, there are large robots that can load large trucks. "But we have never found such a granular distribution on a package delivery vehicle before," says Jarr.

These puzzling aspects may also be attributed to the fact that the distribution centers do not pay much attention to steps beyond their reach, of the interfaces, as Jarr and his project team have discovered. "System providers for warehouses or distribution centers still do not give much thought to what actually happens outside in the next step," notes Jarr. "Conversely, suppliers sometimes still pay too little attention to intralogistics automation.” A lesson for the future, which he considers essential: All systems are networked. A package always comes from somewhere and goes somewhere, regardless of whether it is in the ‘end customer’ business or company deliveries. At this point, Jarr sees a lot of room for further innovations and developments. It is feasible that at some point a packing system could be networked with route planning and could load packages differently onto the transport shelf when the route planner reports a construction site as a potential hurdle to completing the initial route. Or the use of artificial intelligence, which can learn with each additional load. "I believe that in the next 5-10 years, the last mile will change so massively that the classic package delivery model will be turned on its head.”

Part 3 of the "Urban Logistics" series: Package delivery in the city - The STILL LiftRunner and the Linde CiTi truck might be the answers.

Markus Schmermund about solutions that intralogistics may have for the last mile.