The research and development of novel user experience concepts in well-regulated industrial domains face different challenges. Systems in these domains often require backward compatibility and integration with legacy sub-systems and protocols. They must comply with well-defined procedures and standards, and must pass through stringent evaluation processes involving actual users under realistic conditions and scenarios. As a consequence, prototyping and simulations are extensively used. During product development, the level of fidelity of a simulation prototype will directly impact the quality of end-user feedback, minimizing ex-pensive rework of UX in later stages of a project. This paper describes the Distributed Industrial Simulation Platform (DISP), a simulation framework developed within GE that facilitates the rapid prototyping and evaluation of novel industrial UX systems. We present the DISP design and main services showing how it has been used in support of the development and simulation of two UX prototypes in the railroad transportation domain.
Simulations and prototypes are key tools of choice in early stages of industrial product development. In particular, they allow designers to evaluate new user experience (or UX) concepts before they are developed into production systems. The higher the fidelity of prototypes and experiences at this stage, the better the quality of the user feedback collected, and higher are the chances of meeting end-users needs.
The development of high-fidelity simulated user experiences in the industrial domain is non-trivial. The first stage in the development of a new operations concept involves discovery and observational re-search. Frequently, a co-design workshops are conducted with end-users to identify their typical workflow, pain points and possible technology-driven solutions. At this stage, low-fidelity story boarding and wireframe prototypes are used to explore and evaluate several potential solutions and UX concepts (Kim et al. 2017)(Levulis, Kim, and DeLucia 2016). Soon after this stage, there is a need for higher-fidelity system prototyping to validate and refine these concepts. Prototypes allow actual users to experience the techno-logical solutions in situations close to their everyday work setting. The closer the prototype is to the pro-posed solution, the better the design feedback on the concept will be. Two main options are generally avail-able in the production of high-fidelity prototypes: the development of full-fledged simulations, or the rapid prototyping of systems out of existing parts.
Whereas the development of simulations using different of-the-shelf simulation packages is an option (Boer, Bruin, and Verbraeck 2006), they provide poor or no support for the rapid prototyping of domain-specific Human-machine Interfaces (or HMI). Instead, they focus on the abstract modeling and simulation of key aspects of the system underlying controls and interaction. Few of them provide mechanisms to fa-cilitate the integration of legacy systems. Finally, there is poor support for end-user data analytics (Poria et al. 2015), often requiring UX designers to manually gather and analyze multi-modal data from both systems logs and user study sessions (Lahat, Adali, and Jutten 2015).
As a consequence, many UX HMI industrial prototypes opt to design around existing systems, relying on the integration of simulations, with existing subsystems and physical HMI controls. While this improves reuse and provides high-fidelity of HMI controls, it many times comes with high integration costs due to the diversity of protocols and subsystems. In this paper we describe our approach to facilitate the rapid prototyping of UX concepts by discussing the design and implementation of an infrastructure that addresses the main concerns involved in the integration of simulation models with existing industrial subsystems and UI components.
Goals
Railroad network simulator architecture with engineer, remote dispatcher UX researcher
In order to facilitate the rapid prototyping and evaluation of novel UX concepts in industrial domains, we have developed the Distributed Industrial Simulator Platform (DISP) which provides a common component framework and additional services for the development of human- and system-in-the-loop distributed sim-ulations. DISP was developed as part of GE’s Global Research next generation railroad program which includes the use of current HMI technology, advanced networking, controls, and collaboration technology in enhancing today’s railroad operations.
DISP was designed according to the following architectural decisions: Distributed Component Model; Message-driven communication; Common extensible data model; Multi-platform support; and Service-ori-ented extensibility. These design decisions were them implemented in Java, in the form of distributed ser-vices, reusing the communication capability of ZeroMQ middleware (Hintjens 2013). In this section, we will discuss these decisions.