In particular, the terms autonomy and autonomous are used in an attempt to attribute greater value and user benefits to new products with new functions. As there is no generally accepted understanding of the terms automated, fully automated, autonomous, fully autonomous and intelligent in the field of intralogistics, there are a large number of products that are difficult to compare with one another, leading to misunderstandings and disappointed expectations among users.

It is noticeable that the term "autonomous" is used in an inflationary manner, often defined in an idiosyncratic way and used on many manufacturers' websites as a synonym for self-driving, driverless and fully automated. The term "automated guided vehicle system" (AGV system) has been used for more than 60 years and describes a logistics system with which a specific logistics task - for example, transportation to link sources and sinks, assembly lines for series products or a task in warehousing and order picking - is performed by a fleet of automated industrial trucks.

An AGV is an organizational tool that guarantees reliable, safe material transport with maximum performance, availability and quality. The peripherals and all logistics and production processes in the environment are carefully coordinated.

Typical examples are well-planned, complex logistics processes in automotive manufacturing, in automotive supply companies, in logistics centers, at series manufacturers of white and brown goods, in the food industry, flows of goods in hospitals (food, medicines, laundry, waste, etc., away from the wards).

Trend: Focus on vehicles

The vehicles used in such systems are usually referred to as "automated guided vehicles" (AGVs) and can differ greatly in terms of their technological functionalities (mechanical, mechatronic, electric), but also in terms of their "intelligence" (sensors, control functions, autonomy). For some years now, efforts have been underway to develop AGVs that differ from classic AGVs, which are procured as part of a system business.

and realized as a project, to focus on the vehicles and only procure these (product business). These vehicles are often not referred to as AGVs, but as mobile robots (MR), autonomous mobile robots (AMR) or simply "robots". There are also numerous other designations, which are often also product names of individual manufacturers.

The focus is therefore on mobile robots that can be "simply" integrated into an existing industrial environment and can take over simple services (such as transportation, handling, cleaning, information) after a short commissioning period. It is possible for a few such robots to communicate with each other and share tasks. Such vehicles can be used in a variety of ways, require little planning, hardly any preparation of the operating environment and short commissioning times. If necessary, they can function without a stationary AGV control system if they find, distribute and carry out their tasks themselves in coordination with the other MRs.

As the MR and/or AGVs used in such automated guided vehicles are not fundamentally different - although functionality, complexity and intelligence can vary greatly in both cases - the VDI 2510 and 2710 series of guidelines apply equally to both. The guidelines on AGV safety and the AGV safety guidelines (for planners and operators) are also applicable.

There is no either-or

But now to the concept of autonomy: We are not assuming an either-or - AGV or AMR - but are looking at autonomous functions of a system with automated vehicles. We are therefore talking about more or less autonomous vehicles or vehicles with more or less autonomous functions. We limit ourselves here to functions for driving, safety and load handling, whereby it is irrelevant whether the functions are implemented as software locally in the vehicle, in a central control system, in an external cloud or a suitable combination.

In the VDI working group "Autonomy", we have dealt intensively with such functions. We have examined all these functions to determine whether they are "only" automatic functions or "genuine" autonomous functions. Because autonomous functions are complex. As a rule, they involve situational reactions to changing environmental conditions and system states, which are recorded and evaluated using multi-dimensional sensor information. Artificial intelligence methods, such as machine learning, are proven methods for this. However, it is also conceivable that comparable results can be achieved using complex high-level language programming.

A VDI guide will soon be published describing both typical automatic functions and a range of autonomous functions, and explaining why or under what conditions a function belongs in one class or another. The guide is intended to help the reader classify a vehicle or vehicle system in terms of autonomy. They can then assess this solution in terms of its suitability for a specific task. The much-vaunted innovation "autonomous obstacle avoidance" is certainly a productivity-enhancing and therefore necessary function for the cleaning robot in the airport hall - but for an AGV in a production process that is synchronized and designed for maximum efficiency of the transport system, it is rather questionable and not expedient. Conclusion: An autonomous function is not fundamentally good and helpful per se - rather, it must fit the respective application!

About the authors: Thomas Albrecht is Head of Automated Guided Vehicles at the Fraunhofer Institute for Material Flow and Logistics. Günter Ullrich heads the VDI AGV Technical Committee and the AGV Forum. Both are currently working in a VDI working group on a guide to the concept of autonomy in intralogistics.