Flexible interoperability of robots with humans and machines

June, 2023

In modern industrial settings, robots are becoming increasingly prevalent as they offer numerous benefits, such as improved efficiency, precision and safety. However, for robots to reach their full potential, they must be able to operate seamlessly alongside humans and other machines.

In our work, we therefore research standardized interfaces through which humans and other machines can collaborate with robots regardless of the robot’s brand or make. We thereby set the foundation for robots to be seamlessly integrated in smart factories and classical environments dominated by manual labor that can be extended with robots to ease the strain on human workers.

The Need for interoperable Robots

Robots are usually deployed within a larger system, which includes humans and other machines and is designed to fullbring certain types of productive activities. Therefore, from a manufacturer’s point of view, robots cannot be seen as standalone entities, but must be considered in the context of the production system they are part of. Indeed, even a state-of-the-art robot might be detrimental to the overall system if it cannot interact properly with the rest of the system.

When this perspective is taken into account, the interoperability of robots with humans and other machines becomes one of the most relevant considerations when deploying robots in practice as it ensures efficient collaboration between machine and increases the ease with which humans maintain control over the system.

Under the current paradigm, in which robots are produced by many manufacturers from all over the world, robots come with highly diverse user and programming interfaces. This results in challenges when it comes to integrating robots in manufacturing systems. On the one hand, because human experts must be trained for the operation and programming of each type of robot and on the other hand, because specific protocols and data formats must be established to enable machine-to-machine communication between any two systems. While enabling such machine-to-machine interaction is oftentimes relatively easy for machines and robots sourced from one specific supplier, since many suppliers provide integrated ecosystems for their devices, cross-supplier integration remains challenging.

This situation results in challenges for manufacturers who would like to deploy robots in their operations. In particular, they can become highly dependent on a single supplier if they build up non-transferable expertise with the tools and standards created by one specific supplier and thereby get “caught” in its ecosystem. This leaves them not only vulnerable to price increases but also limits their choice when it comes to selecting the most suitable and powerful tools for their operations.

In order to mitigate such undesirable dependencies, manufacturers (and other entities) can create open robot interfaces that can be used to control a large variety of robots and other devices or software programs. Optimally, such interfaces are openly available and widely used. Thereby, manufacturers maintain their independence from suppliers and their ability to flexibly innovate and differentiate from standard market practices.

Our Research on Open Robot Interfaces

Building on top of the network-level integration that is commonly achieved through robots adopting the Internet Protocol (IP) and thereby becoming part of the “Internet of Things”, our research promotes robots to also be part of the Web of Things – that is, we have developed Web-based human- and machine-facing interfaces through which robots can be operated. Very much similar to the Web, which builds on top of the Internet to create an interoperable application layer that enables any user to access any application through one interface (a browser), our Web of Things applications build on top of the World Wide Web and its main protocols (such as HTTP), to create interoperable IoT-based applications accessible through standardized interfaces. These Web-based standardized interfaces dismantle the barriers to interact with and to engineer our robot systems. For example, our system tremendously lowers the barrier for humans to operate robots by providing them a standardized set of commands with which any robot can be operated.

In addition to that, our Web-based systems remain simple to expand and integrate with heterogeneous other soft- and hardware. Thereby, we lay the foundation based on which robots can interact (autonomously) with other machines. This is achieved by exposing robot’s functionalities through so-called “Thing Descriptions” (TD) that have recently been standardized by the World Wide Web consortium where members of our research group are consulted as invited experts. These TDs identify the operations that the robot makes available to other machines or software programs and can be seen as an extension of an API description, such as the widely used OpenAPI descriptions. Thereby these TDs enable other machines or programs to identify a robot and its functionality at run time and to flexibly leverage his capabilities in order to achieve their goals.

By leveraging our technologies, we successfully bypass system boundaries which oftentimes arise when integrating software and hardware components provided by different suppliers. An illustrative example of such a Web-based robotic system that we also make use of in our teaching at the University of St. Gallen can be tested here.

What Lies Ahead: Towards Autonomous Systems

For manufacturers interested in fully leveraging the robots (in combination with all other production factors) deployed in their factories, the interoperability of their robots with other devices as well as humans will become increasingly relevant within the next years. Those manufacturers will have to decide between building up their manufacturing infrastructure based on supplier-specific ecosystems, thereby increasing their dependency on these suppliers, or by building their own infrastructure in order to maintain a higher level of flexibility. The first option can be the easier choice when starting a new project since an ecosystem can provide many features out of the box, the manufacturer becomes dependent on the supplier, which limits the evolvability of the system and can pose a significant constraint, especially in long-term projects, in projects that involve many different suppliers, or in highly competitive and innovative industries, where the suppliers themselves and their products may change a lot over a relatively short amount of time. The second option, on the other hand, comes with a relatively large initial effort but leaves manufacturers with more flexibility and evolvability of their systems.

As shown by our research, the decision to leverage our technologies can serve as a basis based on which intelligent autonomous manufacturing processes can be implemented. This is achieved by leveraging so called software agents that can leverage our standardized robot interfaces to autonomously coordinate the operation of these devices (e.g. autonomous coordination of a welding, a transportation and a packaging robot). The coordination provided by the agents is beneficial to the extent that it increases the ability of the robots to work towards an objective while being able to react to unexpected situations. As a result, agents can be used to monitor and potentially resolve technical failures and enable devices to be changed at run time without needing to update their explicit programming.

With our expertise in this domain, we can advise and support manufacturers in both decision-making as well as building up such future proof infrastructures based on which autonomous systems can be built.