1.2 add a view so that we can zoom up in the sky and look down on the scene to take in the whole track.
Project Objectives: The initial objective of this work is to develop and demonstrate one or more novel railroad information management and display approaches, which may lead to improved productivity and safety for the train engineer and dispatcher functions. With guidance from experienced railroad managers, it is technically feasible to rapidly create a realistic railroad context in which to imagine these displays concepts could function.
Setting the information interface in a realistic railroad context is expected to stimulate joint industry government discussion of likely future operations and safe decision making practices. The team can then rapidly simulate a joint vision of a future operational context. This in turn can be used to make adjustments to the information interface. Later displays and information management software will be used to accelerate ergonomic evaluation of interfaces to proposed advanced train control systems.
Initial Problem Statement: Given a future in which all railroads (trains, tracks, signals, etc.) will be digitally instrumented, this initial task has an opportunity to create a futuristic display environment which will allow today's railroad managers, planners, dispatchers and train engineers to preview future operations and interactively react to unexpected conditions such as unscheduled traffic events or weather constraints.
Strategic Concept
Emergent Temporal Phenomena: Unscheduled traffic events (mechanical breakdown) or reductions in speed due to for example fog often have ramifications beyond the time and place of occurrence. Traffic flowing later in time and remote in space can be affected. The problem is to diagnose the soon to emerge temporal phenomena by for example being able to play out or explore likely events in time and space or zoom out to a bigger picture.
Part of the problem of diagnosing a condition as emergent is in choosing the time frame. Some events such as an accident or thunder storm could from one view seem to occur in a flash but from another view these same events can be observed as a process which emerges over time.
Disruptions to traffic flows may appear to cause a small disturbance in the short term. However, as train speed increases (from 60 to 150 MPH) or service frequency increases from (one train scheduled through to 30) or both; opportunities for a small disruption in the schedule to ripple beyond the local time and place may also increase.
We hope to build a system which allows the dispatcher and planner to visualize, simulate and modulate these phenomena in the context of exploratory usage scenarios.
Tactical Concept
The Journey As each individual train travels its linear path from origin to destination, the journey may be viewed as a series of decision points, or passage through regions of calm or chaos or each step of the journey may illuminate for the traveler possibilities, uncertainties or opportunities. This viewpoint is most often associated with the train engineer. However, most railroad management functions and passengers consider this viewpoint fundamental.
Interface Concept
Global to Local Context: Current railway information displays are unique to the train engineer or local dispatcher or global dispatcher functions, forcing each decision maker to make only certain kinds of decisions and limiting their ability to understand how their separate choices interact. Past experience indicates, separation of operating responsibility lends itself to improved operating efficiency, in the short term. However, separation of information tends to reduce team work and could interfere with progress toward improved safety and operating efficiency, in the long term.
We plan to demonstrate a unifying railroad situation display in which the viewer can continuously change the point of view, spanning temporal histories and zooming geographically from the whole to the particular, aggregating the information units so as to present train operations as the situation demands or as different decision makers require.
Objects of Interest to the Team
The following is a wishlist of the conceptual and physical objects, and processes to model computationally in software.
On a more practical note: the following existing examples are at least worth looking at for inspiration and maybe even for bootstrapping for models and running code.
1. The OpenInventor demo "Slotcar" can be found on the Intergraph NT machines and may come with the SGI version of OpenInventor. We should try to move this over to the Unix platform if not ttthere already and recompile it, or maybe easier, just run it on the PC using the Visual C++ compiler. The geometry here is close enough for early prototypes, and can probably be modified to look reasonably like what we might want to have in a demo of advanced display ideas. First sort of dditions are
2. Mitch Resnick's Starlogo programming environment has an example of creating different kinds of traffic jams, and then solving them. http://www.media.mit.edu/~starlogo/ has been downloaded to the Powermac Rambutan in Users:ronmac. There are many excellent decentralized system examples, the traffic example is particularly useful in that it shows how two rule vehicle systems can produce moving traffic jams. Here traffic comes in from the left. When the rules are; speed up if there is enough space in front, slow down if not, looking just at the car ahead. If traffic starts with each car at a slightly and randomly different speed, soon the traffic jam appears and begins travelling to the left, toward the traffic.
When the look ahead is increased to two cars, the traffic jam gets worse, a totally non intuitive result.
In thinking about ways to visualize the multiple objects of interest, with competing needs to show them in their logical and spatial relations, we came up with a beginning scenario, tilting toward the spatial representation where geographic components are important and we feel we have the luxury of enough display space, with 3D graphics assist to zoom and tilt whenever is needed. The visual idea was to look at the journey along a GIS 2.5D landscape, viewing the landscape from a 45deg angle, vertically separating the layers of interest, letting the layers of most interest to the situation at hand rise to the top of the Z axis, letting the non relevant layers fall to the bottom, with a fixed geographic groundplane (somewhere in there). What would be interesting would be to watch the layers aggregate to form the rule under which the train is currently operating, a form of visual inferencing. Certainly as the train is moving through the landscape the layers will dynamically relocate in terms of their relavance, and this dance should be very interesting to watch.
