Robot Swarms and the Decentralized Future of the Warehouse
The current reliance on centralized system architectures has served the industry well. But distributed and decentralized intelligence, fueled by the Internet of Things and machine learning, is the wave of the future.
14 December 2017
If you haven’t heard it before, the term “robot swarms” may make you a little uncomfortable as you envision an apocalyptic scenario in which robots team up to overthrow humanity.
If so, relax. Robot swarms refers to the ability of a group of robots to work collectively to perform complex tasks, similar to ants in a colony. And, it’s an example of one of the core principles of Industry 4.0: decentralization.
Decentralization, along with Interoperability, Virtualization, Real-time Capabilities, Service Orientation and Modularity, create the foundation for Industry 4.0 solutions in which cyber-physical systems communicate over the Industrial Internet of Things and cooperate with each other and humans in real time.
Centralized system architectures, in which business logic is contained in a central computer system that supports or controls the operation of various subsystems, have been the norm in material handling for years. This architecture can be deployed relatively quickly compared to other architectures and is easier to maintain because all logic is in one location. But, centralized systems have limitations in areas that are becoming increasingly important to warehouse management.
Most notably, they offer limited scalability. A centralized system has a certain capacity and once that capacity is reached, the system can no longer expand to meet growing requirements. They also suffer in the area of fault tolerance: a single failure can take down the whole system.
These factors are driving material-handling technology toward decentralization, either through distributed or fully decentralized architectures.
In a distributed architecture, logic is contained in nodes that support or control remote components or subsystems. A distributed system adapts to its surroundings and propagates data through peer-to-peer communication and forms its functionality and intelligence by combining the capabilities of each node. The Internet, for example, is a distributed system featuring multiple nodes that allow a user to access information across the entire system because all of the nodes work together using a common protocol.
In a fully decentralized architecture, all business logic is embedded in the subsystem or component so that it has all of the intelligence it needs to perform its function, coordinating its activities with other subsystems to handle complex tasks. While intelligence is decentralized, these systems still rely on a central system for component handling, communication and data accumulation.
Decentralization increases the scalability and adaptability of a system. A fully decentralized system is easy to scale—you simply add more subsystems—but will, at some point, reach a limit as the system grows beyond the point where subsystems can effectively communicate with each other. A distributed system offers unlimited scalability. To grow the system, you add more nodes, which only have to communicate with a limited number of related nodes to connect to the entire system.
These decentralized systems also offer the ability to evolve over time and adapt to change—an essential feature in modern material handling equipment, which must adapt to an uncertain future.
So, where do robot swarms come into play? Consider an automation system such as Swisslog’s CarryPick. This system uses small, mobile robots to support goods-to-person picking. Compared to traditional approaches to picking, it offers greater flexibility and productivity. Today, CarryPick features many of advantages of a decentralized system: it is easy to scale the system by adding more robots and it can adapt to changes in warehouse design or even a new facility.
However, CarryPick still depends on the central intelligence and control logic provided by SynQ, the Swisslog WMS. As it evolves, that intelligence will increasingly be migrated to the robots themselves, enabling greater machine-to-machine communication and allowing a “swarm” of CarryPick robots to operate with greater autonomy from the WMS—while still providing the WMS with real-time data on product location— increasing fault tolerance and efficiency.
The vision of Industry 4.0, in which cyber-physical systems cooperate with each other and with the humans they work around, can only be realized when intelligence and control logic is decentralized or distributed across those cyber physical systems. They won’t overtake humanity, but they will enable new levels of productivity, availability and efficiency in the Industry 4.0 warehouse.