By 2030, robots will be central to half of all new warehouses. Humans will only play a role in warehousing if robots don't pan out, predicts research firm Gartner. The robots we deploy then will also be a lot smarter, more flexible and more independent than the current generation. So says consulting firm Capgemini, which talks about 'physical AI'.
Staff for warehousing will remain scarce and expensive, research firm Gartner predicts. And the staff that are available have less and less desire to do manual work. As a result, warehouses will increasingly resort to robots in the coming years. Gartner predicts that in four years' time, half of all new warehouses will be designed around robots.
Fleet of different robots
In warehousing, AI is playing an increasing role. "AI is constantly optimising operations, transforming warehouses from static structures into flexible systems that adapt as demand changes," states Abdil Tunca, senior principal analyst at Gartner. "This is also changing the way companies think about designing warehouses. They are relying less and less on human labour and more and more on solutions that allow them to control an entire fleet of robots."
Because the robot market is fragmented, that fleet will in most cases consist of different types of robots. This calls for a smart platform that allows the warehouse to coordinate the activities of all those different robots. In time, an AI-driven operation may emerge that continuously optimises itself. According to Gartner, human labour will only be needed in warehousing for handling exceptions that the robots cannot cope with.
Physical AI in warehousing
In a comprehensive report called Physical AI, consulting firm Capgemini takes a closer look at this new generation of robotic solutions. Physical AI represents the environment in which modern AI technology is conquering the physical world after the digital world. Robots in this environment will no longer just perform fixed, pre-programmed tasks, but will independently perceive their environment and make independent decisions based on it. Robots can adapt to their environment and perform a variety of tasks.
Capgemini explains the difference between current and next-generation robots. Whereas existing robots often work in controlled, predictable environments, future robots can also cope in messy, constantly changing or even completely new environments. While existing robots often receive instructions from above, future robots can make decisions independently and learn from them. They can understand what people say and thus follow voice commands: 'put these boxes on that pallet'.
Robots for loose loading
As a practical example, Capgemini cites FedEx's pilot with Dexterity, a US robot startup. Within FedEx, it is customary to transport parcels between two hubs loose loaded, i.e. without pallets or other load carriers. This results in a higher load factor, but is a complex task because the parcels vary considerably in weight and dimensions. Dexterity's robots can see which parcels are coming at them in which flight order and quickly calculate the best way to load them without damaging them.
Capgemini also stresses that different robots must work together as one big coordinated system. In this system, each robot is completely autonomous, while digital AI agents monitor all activities and take care of coordination. The role of humans is no longer to meticulously programme all activities, but simply define goals and conditions. The digital AI agents take charge of planning, monitoring progress and intervening if necessary to ensure that all goals are met on time.
Netherlands underestimates potential
Capgemini surveyed almost 1,700 managers from various sectors about the potential of Physical AI. This shows that two-thirds expect this development to be a gamechanger for their sector. When the scores are broken down by country, it turns out that the Netherlands has by far the lowest score. While in Switzerland 77 per cent think Physical AI can be a gamechanger, in the Netherlands the percentage is only 44 per cent. It seems that managers of Dutch companies underestimate the potential of Physical AI for warehousing, among other things.
So what exactly is this potential of Physical AI? Of all those surveyed, 76 per cent expect productivity to increase, 70 per cent expect costs to decrease and 65 per cent expect quality to increase. Other frequently mentioned benefits: continuity of operations (57 per cent), workplace safety (57 per cent) and reduced physical strain (53 per cent). Managers surveyed additionally indicated that Physical AI makes it possible to use robots for tasks that were not previously considered for robotisation.
Humanoids in warehousing
Asked about the reason for investing in Physical AI, 74 per cent point to staff shortages and 69 per cent to rising staff costs. Fear of falling behind the competition is 63 per cent's reason to invest, while 59 per cent cite workplace safety. Most investments go to autonomy mobile robots (60 per cent), followed by industrial robots (56 per cent) and cobots (51 per cent).
Only 23 per cent of those surveyed expect to put most of their money into humanoids. Nevertheless, the majority do see potential in these 'human' robots. The most frequently mentioned benefit (75 per cent) is that humanoids can be deployed in existing work environments designed for humans without having to modify them. The ability to use them for a wide variety of tasks is mentioned by 72 per cent, while 55 per cent say that humanoids can interact with existing tools and machines.
