Don't know enough about the domain, but as I understand the project I would go for a JavaScript based web application fed via websockets with all the information to depict, as long as you don't need a realistic representation of the control room.

You'll may be able to display all that information using some free tolos as d3.js for charts and visualizations; and openlayers or leaflets for the maps.

Dear Antonio,

Thank you for your reply. I am  considering to use Unity3D. Are you, or anyone else, familiar with Unity3D? It would be interesting to know whether others used Unity for workplace / control room simulation.  

Unity3D will allow you a very realistic device-like representation based on a 3d-model of the control room.

However I have no clue how you could integrate it with a geographic information displayer, or with plot libs for charts and other visualizations.

OpenSimulator is another option - you can use VNC to display the video of actual displays on the simulated ones.

Dear Antonio and Mark,

Thank you for your reply. I understood that Unity does not truly support the design of 2D user interfaces in the game. That it can be time consuming to develop common interface items such as menu’s.

Using D3.js, openlayers and leaflets might be a better option, thank you for pointing them out. I, however, also want to be able to easily develop new game-like scenario’s. The idea of having one piece of software to develop the simulation environment and add game-like scenario’s looks attractive. Is there a java based, more 2D focused equivalent of Unity which I could use? Or is the best option really to use libraries like those you pointed out in a java editor?

What are your experiences? Does OpenSimulator support common 2D interface features?

What we have found is that Unity provides a higher fidelity finish, but OpenSimulator has more real-time flexibility.  that said to change a display would either require uploading a new graphic to put on the 3D object or using a VNC connection to display the video of an actual display.  Two things come from this, 1. the display can actually respond to inputs and 2. if you actively change the configuration of the real screen it is reflected in the virtual environment.  Both Unity and OpenSimulator are good products it depends on your use case which to use.  For experimentation and moving things around OpenSimulator has the flexibility, however, I know of cases where people who are familiar with Unity are able to do the same (but it does take more skill/knowledge to use Unity IIRC).

Unity has also 2D capabilities, but it's heavily oriented to action. I mean it probably can help building menues and so, but its real strength is on moving sprites (or 3d model) in huge scenarios with a quick response to input events (keyboard, mouse, etc.). In the end, it's a game engine.

As Mark said OpenSimulator will make easier to change the simulation in real time, letting add events or elements ad-hoc by a coach or tutor while the simulation is running.

I would use either Unity or OpenSimulator if the simulation you want to develop should pay attention to the environment of the control room: walls, tables, people entering and going out, sensors and displays set in different places around the room, etc.

I would avoid them if the simulation is more based on fast data analysis and decision taking. Then I would go for the web interface.

Another option could be Processing or a C++ library like OpenFramework. That tools can help to program a more rich graphical environment, and can take advantage of existing libraries for plotting or geographical información (Java-Python for Processing, C++ for OpenFrameworks).

Do you have a picture/draft on how do you expect the simulation to look like? Maybe that could help us to help you ...

I'm not sure where all of these perceptions about Unity come from, but creating menus (2D and 3D), integrating geographic information, controlling data input/output and just about anything else are very straight-froward in Unity. If you need to connect with other applications, packages, hardware devices and whatnot, you simply use Plugins or sockets. We have created anything from data processing applications to graphically intense simulations of realistic environments and we have found no downside to Unity compared to many other engines. I'd be surprised if Unity is less flexible than OpenSimulator when it comes to changing the simulation and adding events. One of its biggest weaknesses, the 2D interface/menus, has been addressed in the latest release 4.6 which is just about to be officially released. Feel free to contact me if you have further questions about Unity (we're not affiliated with Unity).

Dear Colleagues,

Thank you for your replies. I think I will try Unity to see if it gets me started. Sebastian, thank you for your offer. Maybe you can help me with the following question.

Attached you can find a picture which gives an idea of the workplace which I aim to simulate. The nautical operational network management operators are responsible for incident handling and need to provide relevant information about the situation to skippers, colleagues and emergency services. The workplace includes camera's, radar and AIS data showing the real-time position of ships on a map, visualizations of prognosis information about heavy traffic and obstructions, information about active and planned traffic measures, detailed information about ships and water management objects, etc.

I redesigned parts of the interface for my study. Some parts, like the window displaying information about ships, will remain the same compared to a different, real-life study. In this study, we programmed these windows using JavaScript with AngularJS as framework. The server part is written in Java. Will I be able to reuse the JavaScript of the real-life study in Unity?

I am not aware of any workflow that enables you to implement AngularJS within Unity. But you might be better off posting this question in the Unity forums, as someone might have been in a similar situation.

Looking at the screenshot, my main question is: are the data displayed on these maps and interfaces actually live data or are all of these recorded or mockup datasets that you are using for your studies?  You mentioned a server so I assume you are querying the data from somewhere? If that is not the case, I don't see much of an issue with re-creating the interface within Unity. That being said, there are plenty of options to also implement a server infrastructure to get live data into Unity. I have the feeling this might be better suited for an offline conversation.

I don't have much to contribute to all that has been said so far but perhaps it would prove advantageous to pinpoint what kind of actions do user would have to do in-game. Are they meant to be be real time or can it be more turn based or event driven?

I'm saying this because using a simple tool such as ren'py (http://www.renpy.org/) could easily decrease development time by factors of 3 and 4. Also pygame (http://www.pygame.org/news.html)  is pretty accessible if you need more flexibility.

Evidently some game engines are very genre oriented. This means less development time but also less flexibility. But perhaps for the type of game you're planning you don't need  to use something as powerful as Unity.

Nowadays it's easy to find lists of game engines in the web. Sometimes on those lists a real jewel appears. (fast google example search: http://www.slant.co/topics/341/~what-are-the-best-2d-game-engines)

That said, as far as I know, If you have the time to develop in Unity than do so. It's probably one of the best game engines today and it will allow you to do plenty. 

Considering 2D and web-hosted development you may want to look at http://forio.com. For 3D you may like to check this out: http://universalcampus.igb.uci.edu/ (similar to secondlife) and move forward from here. They provide free resources. 

I would say ren'py is too genre-oriented for this kind of simulation. Meanwhile pygame belongs to the same set of languages/libraries than Processing or OpenFrameworks.

I do work with a nautical simulator and our system is quite complex and large, currently composed by 7 different modules. We are highly focused on visual fidelity and mathematical precision (regarding the ship's behavior). We also have a Radar simulator and several Instruments, which are heavily GUI-based. For the most part, we use Unity and it fits us pretty well. FYI, our Instructor module was built using Qt and C++.

If you're focused solely on user interface (GUI) then I believe Unity (as of the current 4.5 version) is not so friendly. They will release a new GUI system very soon but developing UI elements with the current version can be very disappointing. Hope I could help!

Ellemieke, this is a rather complicated issue and I will try to answer as concisely as possible.

I think the big question here is the level of functional and/or structural fidelity required. This will be determined by what results you hope to obtain from the evaluation. If the physical environment is important (i.e. ecological validity) then you will require a physical mockup of the control room with all the physical artifacts (consoles, instruments, controls, displays, etc.) in the intended locations. The functional validity is far more involved and will require sophisticated software that can simulate every object users will interact with in the course of a specific task. It is possible to build a part-task simulator if it is not necessary to evaluate all possible operating scenarios.

At Idaho National Laboratory (INL) we have a large facility called the Human System Simulation Laboratory (HSSL) where we can simulate an entire nuclear power plant control room. This facility has 15 consoles, each with three 42 inch displays in a vertical arrangement. The bottom two are touch screens and this arrangement simulates the actual layout of a control room with all instruments and controls digitally simulated. All objects are active and respond to operator interaction in exactly the same way as in the real control room. This is called a full-scope functional simulator and it is exactly the same as the training simulators at power plants that are used to train reactor operators. The emphasis in the HSSL is on functional validity and we use it in research on operator interaction with new designs for control room upgrades, etc. The HSSL is also equipped with a suite of physiological and psychological measurement equipment that we can use to assess situation awareness, stress, workload, etc.

The HSSL is also a reconfigurable simulator - we can change the physical layout of the consoles to match the layout of different control rooms and we can load different software simulations to match different reactor designs at different power plants. This software is probably more complex than what is needed for nautical traffic control simulation. This is because the physics of all systems and processes in the power plant are accurately simulated. Nevertheless, as Giancarlo in the previous comment pointed out, building any kind of simulator that matches real-world functions is not a trivial undertaking. The software that we use in the HSSL is developed by companies that specialize in this kind of thing (e.g. GSE Systems, Western Services Corporation, Honeywell, etc.) For example, building a nuclear power plant simulator from scratch takes about 8 years and costs millions of dollars. It is possible to build the kind of software simulator you have in mind using commercial software like ABB System 800xA Process Portal (this is the front end of software typically used to develop a distributed control system). Other options are Wonderware, Simatic Win CC, LabView, MatLab, etc.

So, as I said in the beginning, it all depends on the quality and validity of results required, which will in turn determine the level of fidelity required in your simulator. I you have an existing task analysis and human performance requirements for this environment, it would help determine the nature of the simulator required.

Finally, it is also possible to use discrete event simulation tools like Arena or IPME (Integrated Performance Modeling Environment) to build a part-task simulation of specific complex tasks. This kind of simulation is cheap and relatively easy to develop, but can provide very accurate results, especially on workload.

Hope this helps!

Dear Colleagues,

Thank you all for your full and helpful answers to my question. At this moment, I have almost a year to build a nautical operational network management simulator with the help of a small development team. The aim is to build a reasonably realistic simulator, in order to be able to study the effect of certain user interface and prognosis features on operators’ situation awareness. The user actions are meant to be event driven, as we aim to study operators while they have to deal with incident situations and extreme weather conditions. At this moment, we are already developing a pilot environment which we will use for real-life testing. Due to safety risks, we however cannot use the real-life environment to test risky situations. Summarizing all your contributions, I decided to start using Unity and see how far we can get in a month time. I hope we are able to reuse some of the software which we developed for our real-life pilot environment. Maybe we can use the GUI which we developed there, but use Unity to feed data and to simulate camera images. Do you have experience with such a combination of using Unity together with an external GUI? Our pilot GUI is build in JavaScript with AngularJS as framework. After a month, I’ll decide whether to continue with Unity. If I do not believe that it will allow me to develop what we need in a year time, then I will probably rethink this decision.

Unity 3D is an efficient widespread and multi platform soft for creating games and simulation. Most interfaces and data tools already exist and are available within the user's community

An update on my progress for those of you who followed this questions because you deal with similar research:

Two programmers started to build the nautical traffic management control room simulator. We use Unity to build the simulation part, and separately build a front end which communicates with Unity. In a month time, among other things we were able to:

- build the waterway network with the use of existing ENC maps

- included path finding on a part of the network, where vessels automatically follow the most logical route

- include locks which we can put into maintenance to cause hinder

- developed the basics of the GUI for an object planner role and operational network management role and included communication between the two operator roles and the simulated vessels

- after the operator sends a notice to skippers, the vessels in the network adapt their route accordingly

At this moment, we include the ability to set or release traffic measures such as speed limits on a part of the waterway.

We concluded that the use of Unity indeed helps to rather quickly develop a traffic management simulator. Thank you all for your advice and good luck with you future work!