APS.net
This SFP investigates models, architectures, techniques, and algorithms for increasing the flexibility and adaptability of industrial production systems. Software, and specifically, software architecture plays a central role in achieving these goals. The general capabilities of a production plant depend on its physical layout. Yet, which capabilities are invoked, in which order and under which conditions is controlled mostly by software or human operators. Thus, fast and cheap reconfiguration can only happen through software designed to allow for adaptability and flexibility. In these systems, physical aspects such as material flow, manipulation of physical objects, and physical layout of machines and humans, play a major role.
In this SFP, we borrow concepts, approaches, and ideas from software Architecture to guide the design of Cyber-Physical Production Systems (CPPS). Adaptability in CPPS comes in two main categories: adaptation of the software (i.e., machine configuration, process configuration etc.) and adaptation of the physical layout (i.e., relocating machine, mobile robots, autonomous guided vehicles). Both categories imply software adaptability.
Traditionally, with little or no product change, engineers custom tailor the software for the machines/robots/production cells specifically for a particular product. With increasing demand for adaptability, two orthogonal adaptation dimensions emerge. On the one hand, we distinguish between the levels of adaptation, and on the other hand, we differentiate according to the locality of adaptation. The former describes adaptation of product-specific vs machine-specific code, while the latter separates adaptation within a machine, invisible to the outside (local), from adaptations affecting multiple machines (distributed).
APS.net investigates models for achieving interoperability on multiple levels. We aim to achieve this by ensuring such a model will allow hierarchical/self-similar modeling of shop floors down to individual software components within a machine. From the point of view of a single component, interoperability can then occur on the same level as well as with components on lower levels and higher levels, while exposing capabilities, allowing discovery and monitoring regardless of hierarchy.
Such a model allows to define blue prints for (i) which capabilities are needed in a production processes, (ii) describe collaboration among production cells, machines, robots – hence supporting the cognitive reasoning on a component’s surroundings and its role within, (iii) allow reasoning on optimally distribute control and dataflow, for (iv) ultimately achieve distributed process execution.
Goals
The overall applied research-centric goal is to investigate a new middleware for the shop floor that enables semantic interoperability and flexible adaptation of machines and shop floor configuration. In particular, the focus is on the question of how machines, robots, and increasing demand for adaptability, two orthogonal adaptation dimensions emerge. On the one hand, we distinguish between the levels of adaptation, and on the other hand, we differentiate according to the locality of adaptation. The former describes adaptation technical interoperability through interface standards, (ii) achieving semantic interoperability through the use of data standards, (iii) support for programmatic interoperability through infrastructure & central services, and (iv) support of engineering, development, deployment, operations of modular and adaptive systems. The ultimate goal is having a framework that allows the discovery of production entities, composition of their capabilities, distribution for decentralized execution, monitoring of that execution, and continuous adaptation thereof.
Approach
The approach is based on “Design Thinking”. Stage1 Empathize: Interviews and workshops with company partners showed the current limits of flexibility of machines and processes at the shop floor. Stage 2 Definition: Based on this, specific objectives of flexibilisati on were defined and use cases were developed in which target attainment was to be measured (e.g. relocation of a production process from one production cell to a non-identical one). Stage 3 Ideate: Based on a study of the state of the art and research, several architectural solutions have been identified. Stage 4 Prototype: Simple/ advanced prototypes focused on the basics and iterative enhanced prototypes allow fast implementation of ideas. A first simulation of flexible machines and OPC-UA was realised. Stage 5 test: Scientists and engineers at the industrial partner evaluate the prototype and thereby generate feedback for the previous four stages. The whole process is highly iterative and non-linear; feedback from each stage to previous ones is not only possible but also explicitly desired.
Expected and Achieved Results
This project has two main expected results: (i) a framework for modeling capability-based production processes, and (ii) a framework for distributed/adaptive execution of production processes. Along the lines of the former expected result, an extensible meta-model provides the basis for modeling actors (humans, machines, robots), processes (work steps including control and data fl ow), parts (the physically input and output of work steps), and resources (such as tools). A key aspect are “capabilities” which describe abilities that humans, robots, machines provide without having a tight coupling to the providing actor. A first version of such a meta-model is complete. This model serves as the basis for an algorithm to semi-automati cally match discovered capabilities (from machines etc.) to abstract processes (i.e., based on capabilities only).
Model and algorithm are available in an editor. The second main expected results, where preliminary aspects are complete, is an algorithm for analysing the control and data flow among process steps to allow optimally allocating not only capabilities but also control logic to actors in a distributed fashion, thus enabling decentralized process execution. The distribution procedure involves directly linking up actors that need close collaboration such as machine-robot synchronized actions, dynamically generating and deploying code, as well as dynamically interpreting and executing subprocesses on machines and robots. Production step distribution and execution is only one aspect. Scheduling multiple process across the same machine is an equally relevant, orthogonal challenge. The expected result is creating a scheduling algorithm (based on prior project results) that considers failing transport mechanisms (e.g., AGVs) as well as machine failure likelihood to produce resilient plans. Such resilient plans may be less optimal in terms of throughput, but require less costly, less impactful, changes in the event of failures, which is especially relevant during ghost shifts.
