Swarm intelligence navigation

Swarm intelligence in warehousing

A truck that has the ability to plan autonomously is a visionary idea for intralogistics. But is automation only conceivable in combination with swarm intelligence navigation?



An ant has no sense of geometry whatsoever, yet ant trails nearly always follow a geometric pattern of efficiently planned straight lines. The ant algorithm is an example from nature where the cooperation of a group brings about a result that would not have been possible by an individual. In this example, the ants initially move randomly but then systematically as scents are used to identify the shortest route.

Admittedly a warehouse is not an ant colony, but would it not be possible for trucks and goods to independently plan their routes in this way? This visionary idea also carries the hope that swarm intelligence could offer tangible benefits in terms of resilience and flexibility without the need for a central control system. This is of particular interest because fast-moving operations are vital in warehouse logistics and, at the same time, there is an increasing tendency towards customised packaging sizes and conditions are constantly changing. What we are looking for is a logistics system that is both variable and efficient.

What if the truck were capable of autonomous planning…

"Companies that are interested in manufacturing automated trucks are essentially having to invest twice: in the truck itself and in the integration of the truck into the processes and warehouse environment," says Joachim Tödter, Head of Technology and Innovation at the KION Group. The idea of a truck that independently dials into the Wi-Fi network as soon as it starts operating and is able to coordinate its activities with all the other trucks in the warehouse is certainly compelling. But before this vision can become a reality, there are a number of questions that need to be answered.

Firstly, the issue of autonomous navigation within a space. On the shopfloor, trucks frequently encounter all kinds of situations: objects obstructing their path, employees walking around, and racking being reconfigured to create completely new spaces. But conventional trucks follow specified routes. Isolated attempts have been made to change this. Linde is working together with its French partner Balyo to develop trucks under the 'Linde Robotics' label that can move independently around the warehouse with the help of geo-navigation and 3D cameras.

STILL has launched the iGo neo, which can autonomously move along aisles in the warehouse. "It may sound banal, but it is a completely new approach for intralogistics," says Tödter. There are a number of indications that the boundaries between manual and automatic control will become ever more blurred and that trucks will operate in a way that combines both in future. In today's environment, greater flexibility is often more important for logistics planners and warehouse managers than the prospect of fully automated navigation.

Complex interaction

The other key point is the interaction between different vehicles. A great deal of research is already under way in this area. STILL has combined shuttle vehicles and industrial trucks in its Hub2Move project. In the collaborative project 'marion', a tow tractor and a reach truck have directly interacted with each other without being given instructions from a central unit. Operations such as these require certain algorithms, explains Tödter. "The trucks have to negotiate where they will meet and the exact point in time at which they will transfer, drop off or collect the consignment." When automated industrial trucks can master these tasks, it will be the first step towards swarm intelligence. But in view of the actual warehousing environment, it would be more appropriate to also include manned trucks that would move around the warehouse on an equal footing with fully automated trucks. The more complex the interaction, the more complex the algorithms and the requirements – not least the demands on the network. A latency period in the Wi-Fi network of a single second could lead to an accident between two trucks that were moving towards each other. Here, hopes are resting on further developments in high-speed networks, the 5G network, which could set new standards in data transfer.

The benefits of a combined system

But when all these questions have been resolved and trucks are able to independently move around and interact with each other, Tödter asks whether the additional step of fully decentralised swarm intelligence is really necessary. "We could have central control over operational planning but let the trucks negotiate certain situations, such as which truck has to give way to another truck," he says. The only tangible benefit that a decentralised system could still offer would probably be the low outage risk – using a structure that would show redundancies both in the transmission paths and in the orders being carried out, so that a single outage would not cause the whole complex to come to a standstill. Apart from that, a combined system using both central instructions and swarm intelligence would have far more advantages," says Tödter.

"The research into this is on the right track," he says. And with its many projects, the KION Group is at the forefront of this research. "The most important thing for customers is being flexible and minimising the work and expense involved in putting the trucks into operation," Tödter concludes. Swarm intelligence is one conceivable way of doing that – but it will be far more important to improve the capabilities of individual trucks overall.