Are Robots Taking Over our Warehouses?
Updated 21th January 2022
There has been much talk of robots taking the place of humans in warehouses for many years. In fact, some commentators may argue that, with the advent of timed processes, labour management systems, performance monitoring and radio data terminals, many warehouse operatives are already expected to work like robots. Thus real robots may well be much better suited to many of the current warehouse tasks – with humans planning their deployment in accordance with constantly varying demand patterns.
Automation has of course been fairly widespread in warehouses for many decades now, covering such areas as:
- Automated storage and retrieval systems for the storage of pallets
- Miniload systems for the storage of cartons and tote bins
- Automated guided vehicles (AGVs) for the movement of goods
- A-frame dispensers for order picking small cartons
- Advanced conveyor and high-speed sortation systems
- Automated packing machines, pallet loaders, and vehicle loading systems
However, can any of these actually be classified as being a robot? An industrial robot is defined by ISO 8373 as:
An automatically controlled, reprogrammable, multipurpose manipulator programmable in three or more axes, which may be either fixed in place or mobile for use in industrial automation applications.(1)
It could be argued that all of the equipment types mentioned above are automatically controlled and reprogrammable. However, some are not programmable in three or more axes and it is doubtful whether any are truly multipurpose.
More recently, there have been a number of new equipment types that have been introduced under the name of "robots".
One that has gained quite a lot of publicity is the shelf modules-to-picker system using robotic drive units (also known as ‘bots’ or robotic butlers) to bring shelf modules (say about one metre in length and comprising about two to five shelves in height) to a picking station for manual picking. Amazon, for example, has adopted this type of system in a number of their U.S. warehouses.
Another is based on a multi-level grid system whereby tote bins are stored in vertical columns in a grid frame and accessed by multi-wheeled robots running along the top of the frame. These extract the required tote bins and take them to ports at the edge of the grid for onward movement to a picking station.
There are also robotic fork-lift trucks which can put away and extract pallets to and from racking. These are similar in many ways to AGVs but also have a lifting capability. Similar trucks are used to transport pallets along pick faces so that order pickers do not have to drive picking trucks during a pick route.
Examples of robots with arms include robotic picking machines which use grippers or suction pads to pick up items. Improvements in 3-D vision systems are making these technologies more feasible. As with the options of picker-to-goods or vice versa in manual solutions, there is the decision of robot-to-goods or goods-to-robot with this equipment.
A similar application found in warehouses is that of robotic palletisers which can stack tote bins onto wooden pallets or onto dollies (wheeled platforms).
Even though these are all called robots, it is doubtful whether they all fully meet the ISO 8373 definition, particularly with regard to being ‘multipurpose manipulators’ - with the possible exception of some robotic picking machines and palletizers. However, most people probably use a similar definition to one that I found on the internet from a computer 'guru’:
I can’t really define robot. I just know one when I see it.’(2)
Thus, the answer as to whether or not robots are taking over our warehouses is probably in the eye of the beholder.
UPDATE (2018): Are "Cobots" Collaborating with us in the Warehouse?
Since writing the above about robots taking over our warehouses in 2016, there has been a great deal of interest in warehouse robotics and a large number of new developments in the field. Also, much of the emphasis has changed from robots taking over the jobs of warehouse workers to “cobots” collaborating with warehouse workers.
In order to understand this shift in emphasis, it is necessary to look at some of the advances in the technologies that robots use. These technologies include:
- Advanced location and navigation systems such as LiDAR (Light Detection And Ranging) providing information for SLAM (Simultaneous Localisation and Mapping). These know where they are in a warehouse and can navigate the best way around, even with changing situations such as blocked routes and congested aisles.
- New end effector solutions (e.g. advanced gripper and suction devices) to enable robots to pick up items more effectively.
- Modularisation enabling additional tools to be added easily to robots to increase the variety of warehouse tasks that they can perform.
- Intelligent vision systems, sensors and 3D cameras; so that robots can, for example, see and recognise items in picking locations or see fork pockets in pallets for fork-lift type operations.
- AI (Artificial Intelligence) and machine learning; robots learning the best way to pick up an item when it is oriented in various positions in a picking location.
- VR (Virtual Reality) systems. A person wearing a VR system can teach a robot that basic movements of picking an item and this can then be enhanced by the robot learning how to pick in changing situations through AI.
- Cloud technology (i.e. processing and storing data on remote servers) enables manufacturers to monitor robots (e.g. for breakdowns) in real-time or take over if one has encountered a situation it cannot resolve.
- Industry 4.0, Big Data and the IoT (Internet of Things). Robots can read individual item identifications and update databases through the internet as the status of items is changed. E.g. Item now picked or item outside temperature control limits.
- RPA (Robotic Process Automation) enables the automatic mapping of processes, for example, to link warehouse robot and worker operations to the wider supply chain.
Whilst the basic types of robot have not changed greatly, the technologies that enable robot operations have become far more sophisticated. As these technologies become more widespread (both in the field of robots and in many other fields) their costs are decreasing. Together with warehouse staff recruitment difficulties in many advanced economies, this has led to the investment payback for robots becoming much more attractive.
It was only a few years ago that most industrial robots operated in a caged area away from people. However, the nature of recent technological advances, for example in sensing and navigation, has led to the increasing suitability of robots to work alongside warehouse workers. Examples of “cobots” that work collaboratively with people are robotic drive units that bring shelf units to a manual picking station, autonomous mobile robots (AMRs) that take picked trollies to the dispatch area, and robotic picking trucks (typically also AMRs) that accompany order pickers on their picking run. These latter trucks often include visual displays and put lights (to help locate the items in the correct position on the truck). It may be the case that the robotic truck actually picks the straightforward items and the person picks the more unusual items. The robot could then learn how to pick these unusual items through the use of virtual reality and artificial intelligence.
These trends lead to the question as to whether “cobots” represent the future of warehousing or whether they are merely a step towards full robotisation. The answer will almost certainly depend on the nature of future technological advances, the future costs of robots and the future costs/availability of warehouse staff.
Whatever happens, there will still be a need for people to program, control and maintain the robots – or will there? That is a much wider and more difficult question!
UPDATE (2022): What can Robots Actually do in a Warehouse?
There are a multitude of ‘robots’ available from materials handling suppliers for use in warehouses, offering benefits such as labor savings, faster throughput and greater accuracy. However, are all of these in fact robots and what can they actually do?
The International Federation for Robotics (IFR) adopts the definition set out by the International Organization for Standardization (ISO) in that an industrial robot is an “automatically controlled, reprogrammable, multipurpose manipulator, programmable in three or more axes, which can be either fixed in place or mobile”. This definition tends to be used to describe robots with arms that have two or more joints – but not all warehouse ‘robots’ fit this description. In this article, a slightly looser definition is therefore used – being basically whatever is described by materials handling suppliers and the warehousing trade journals as ‘robots’. A common theme of these ‘robots’ is that they have some degree of autonomy in terms of making their own decisions rather than relying entirely on a central computer system.
A common form of robot is one that can move around the warehouse by itself. Although these may look like automated guided vehicles (AGVs), they are ‘autonomous’ rather than ‘guided’ and can therefore find their own way around obstacles rather than being guided on a set path by, for example, wires buried in the warehouse floor. They normally do this by using cameras and lasers which feed data into light detecting and ranging (LiDAR) systems, linked to simultaneous localization and mapping (SLAM) software, which can create a map of an unknown environment and simultaneously recognize its own position on that map. In this way, AMRs can refer to a digital map of the warehouse and find their way around congested areas, employing algorithms and artificial intelligence (AI). Typical applications are:
- Autonomous mobile robots (AMRs) used for transporting pallets, cases, or other items from and to various locations in the warehouse.
- ‘Robotic drive units’, also known as ‘bots’ or ‘robotic butlers’, that bring whole shelf units to a picking station for a person to extract the required goods and then return the shelf unit back to its location.
- ‘Cobots’, also known as ‘collaborative AMRs’ or ‘chucks’, that can lead human order pickers on the optimum route through the warehouse and may provide instructions on a screen of what and how many items to pick, showing photographs of the required items and lighting the pick locations as required. They may also incorporate ‘put-to-light’ technology indicating to the picker into which tote bin, or carton, to place the goods. The ‘cobots’ would then take the filled tote bins to the packing, or dispatch, area. While this is occurring a further ‘cobot’ may then assist the order picker.
When combined with cameras and vision technology on lift mechanisms, then the AMRs can work in the upper levels of racking, thus acting as:
- Autonomous lift trucks (ie counterbalanced fork-lift trucks, reach trucks, or narrow-aisle trucks) that can navigate their way around a warehouse and then put away, and retrieve, pallets in the upper levels of racking.
- Stock-taking robots, which can move along the aisles of warehouses (particularly when the warehouse is closed) in order to check what goods are in each location. These robots may have long vertical masts with cameras at the end so that they can recognize items (or bar codes of items) at high levels in the racking or shelving.
Robots may also be fitted with various handling attachments to perform a wide range of tasks, including:
- Case picking robots that extract cases from shelving by means of clamps or suction pads. Thus, these robots can find their own way to shelving locations and, with the assistance of for example clamps raised on a scissor lift, may pick a case and take it to dispatch or take it to a manual order-picking station.
- Miniload storage systems can be served by free-ranging robots (instead of cranes). These robots can work in the aisles both horizontally and vertically, thus putting away and extracting tote bins at any location in that aisle. The robots can also move into different aisles and then move across the warehouse floor to transport tote bins of goods to manual picking stations – even presenting at the ideal picking height by going up a slope or lifting the tote bin with an integral scissor-lift.
- Robots with hoist mechanisms can serve multi-level grid systems where tote bins are stored one on top of another in a grid frame. The tote bins are inserted and extracted from the top by robots that run on wheels along the top of the grid. They lift the required tote bin or, if there are other tote bins on top of the one required, they would extract these first and relocate them into other columns of the grid frame. The required bin is then transferred to a pick station at the side of the grid frame and, finally, the robots would return the tote bin into an appropriate location in the grid after manual picking has taken place.
- AMR sorters can be used in the same way as conveyorized sortation systems to sort goods to customer orders after picking. These sorters work in groups (or ‘swarms’) and pick up items and deposit them in the appropriate chutes. They are normally fitted with a tilt-tray or cross-belt mechanism similar to automated sorters.
- AMRs fitted with scrubbers or sweepers act as robotic floor cleaners that find their way around a warehouse in a similar way to robotic lawnmowers used in gardens.
Most people think of industrial robots as having an arm (or arms) with two or three joints, and this view matches the ISO definition more closely. There are many examples of such robots in warehouses, including:
- Static robotic pickers which actually pick the goods themselves rather than assisting human pickers, as in the case of most of the examples mentioned so far. These robotic pickers normally have jointed arms with tools such as grippers on the ends. They recognize the orientation of the goods, decide how to grip the required item and then extract it and place in a bin or carton for the appropriate customer order. The technology used to complete these tasks includes camera technology, neural networks and machine learning / deep learning - the latter being forms of artificial intelligence (AI). In such goods-to-robot systems, the required tote bins can be brought to the robotic pickers from mini-load storage by means of conveyors (or indeed free-ranging AMRs).
- Robotic put-walls are similar except that each customer order is represented by a bin – which is normally arranged in a circular multi-level fashion around the robot so that the robot can easily reach each bin with its joint arm.
- Mobile piece-picking robots combine the jointed arm/gripper features of static robotic pickers with the autonomous movement abilities of AMRs. In this way, mobile piece-picking robots can traverse the warehouse to visit pick locations and extract the relevant items required for a customer order (or orders).
- Robotic case unloading systems can be used to extract cartons that have been loose loaded into shipping containers, using similar technology. This type of equipment can enter a container and unload the cartons using a robotic arm with grippers or suction pads. The goods may then be placed by the robot on an extendible conveyor to transfer the goods to the storage area or wherever they are next needed.
- Robotic palletizers also normally have jointed arms and grippers or suction pads. These can stack cases onto pallets in the most effective way to make the best use of space, avoid crushing sensitive items and achieve a stable load.
- Similarly, robotic tote bin loaders can be used to load tote bins onto dollies (ie wheeled platforms used to transport tote bins to shops).
Robots are thus now being used in almost every aspect of warehouse operations. Whereas traditional automation normally requires major capital investment and lengthy implementation timescales, robots can often be added in an incremental fashion and thus avoid these difficulties. In addition, there is a wide range of options as to the degree of reliance on robots – for example, they can be used as ‘cobots’ in collaboration with human pickers through to fully robotic piece picking AMRs that traverse the warehouse and pick the goods themselves.
Whilst ‘swarms’ of robots controlled by artificial intelligence may not be a common sight in warehouses today, these may well become the norm in the future. This obviously has significant implications for warehouse staffing levels and recruitment needs, as well as for future information technology, maintenance and operational management methods – particularly as deep learning and other forms of artificial intelligence are developed further.
Further details on the full range of options for warehouse operations, including manual, automated and robotic solutions, can be found in The Handbook of Logistics and Distribution Management, 7th Edition.
(1) ISO: International Organization for Standardization.
(2) Gordon McComb (amateur robotics guru): http://www.bowlesphysics.com/images/Robotics_-_A_historical_perspective.pdf