University of California, Davis - ProtoVis
VAST 2008 Challenge
Authors and Affiliations:
Carlos D. Correa
Tarik Crnovrsanin
James Shearer
Christopher Muelder
Ryan Armstrong
Kwan-Liu Ma, ma@cs.ucdavis.edu [PRIMARY contact]
University of California, Davis
Tool:
We built two prototypes using Processing, an open-source programming language and environment for animation and interaction. The first prototype is a 2D interactive visualization of the evacuation traces, where we can track different actors on a sketch representation of the building. The second tool is a timeline, where we visualize the entire evacuation procedure in a single image. It plots distance to the bomb vs. time for each of the actors, and allows us to highlight different timelines of interest, such as possible casualties, suspects and witnesses.
More information about processing:
HTTP
| Processing
Video:
WMV (33 MB)
Traces-1 Where was the device set off?
Grid cell number of where the device went off :
66 x 33
Short Answer:
To visualize the grid cell where the device went off, we created an interactive animation that plots the positions of all actors over time. We automatically labeled as casualties those actors that stopped moving for a given amount of time. By tracking down the first casualties, we narrowed down the position to the cell 66 x 33. We also noticed a suspect stopping for a while at that position. We identified him as Ramon Katalanow, which connects him to the Catalano family, albeit the slight change in spelling.
|
| Figure 1. Prototype interactive animation of evacuation traces. Based on the behavior of actor 21 (yellow) and the location of the first casualties, we believe the bomb was set off in cell 66 x 33, as indicated by the crosshair. |
Traces-2 Identify potential suspects and/or witnesses to the event
List of RFID tag numbers :
1,2,18,21,27,28, 29,30,44, 50,56,76,81
Short Answer:
After localizing the cell where the device was set off, we use a visualization where we plot the distance to that cell vs. time, as depicted in Fig.2. We can see from these 5 actors of interest. One of them, RFID 21, shown as a yellow line, moves towards the cell and leaves, managing to escape. We believe it corresponds to the bomber. Other lines appear to move before the incident, which we immediately tag as suspects, RFIDs 28, 29, 44 and 56. Having detected the main suspect (21), we tagged every actor that had line of sight with him as a possible witness, marked as green lines (1,2, 18, 27, 30, 50, 76 and 81).
|
| Figure 2. Timeline Visualization, where we plot distance to the bomb (Y-axis) vs. Time (X-axis) for each of the actors. We noticed different types of behaviors, which are color coded as follows: Yellow is the main suspect as the bomber, since casualties begin occurring around the point where he paused. People who moved prior to the bomb are color-coded in cyan as suspects. People who stopped moving for a while are tagged as casualties and marked with an 'X'. The rest are coded as red lines. |
Traces-3 Identify any suspects and/or witnesses who managed to escape the building.
List of RFID tag numbers :
1,2,21,27,28, 29,30,44,81
Short Answer:
By detecting the casualties and marking them clearly, we are able to visualize those suspects and witnesses that managed to escape the building. These are shown Fig. 2 as cyan and green lines that extend towards the end of the timeline. Suspects that manage to escape are 21 (main bomber), 28, 29 and 44. The others (1,2,27,30 and 81) are those witnesses that had line of sight with the bomber and therefore can provide important information to the authorities. We validated our results with our interactive animation.
Traces-4 Identify any casualties.
List of RFID tag numbers :
18, 19, 36, 39, 47, 50, 56, 59, 60, 65, 69, 76, 78
Short Answer
To quickly visualize the casualties, we detected those tags that stopped moving for a given amount of time after the incident. These are shown in the timeline (Fig. 2) with an X at the end of their life line. Looking at the moment in time where they died, we can also identify some of the deaths following the explosion (18,19,50,56 and 76), and others due to fire or smoke inhalation, after an attempt to escape (39, 47, 59, 60, 65, 69, 78). Person 36 never moves during the entire period. Figure 3 also shows the result of the casualties on the actual map, and as a summary on the right.
|
| Figure 3. Final stage of the animation. We used this to validate our claims from the timeline visualization, which provides cues for the casualties in a single image. |
Traces-5 Describe the evacuation
Detailed Answer:
The ability to depict the evacuation procedure as a timeline, as shown in Fig. 2, gives us an overview of the event in a single snapshot. The concentration of live lines at the end of the timeline denotes a rough estimate of the exit areas. We can form a hypothesis of the evacuation procedure by just looking at that image. Most people escaped through the exit on the south west corner, and gathered around two main areas, shown as the clusters in the upper part of the time line, shown in Fig.4 and labeled as places (a) and (b). There are other people who evacuated to that same area, including a suspect and two witnesses. They made it out through a different door, though (the south east exit). The majority of people who evacuated from the south east exit gathered around region (c). People shown in regions (d) and (e) are those who evacuated from the northeast exit. One in particular evacuated towards a different region than the others (e). Finally, only two people escaped through the eastern exit (f), including the main suspect, Ramon Katalanow.
|
| Figure 4. Overall distribution of evacuation areas. Most people evacuated through the southern exits, farthest away from the bomb (upper lines in the timeline). |
People close to the southeast corner were also able to escape. The fire after the explosion seems to move more quickly towards the northeast exit, which explains a lot of casualties near that exit. We can see that, while people were still trying to evacuate on the southwest exit (the clusters on top of the timeline), the people who did not manage to escape by that time on the northwest corner (Fig.4(d)), were already dead.
If we look again at the timeline and select some of the casualties near the northwest exit, as shown in Fig.5 in the grayed areas, we see that these people were quite far from the exit, which suggests that they were either lost (probably due to panic) or not aware of the evacuation procedures. The people who evacuated through their closest exits manage, in most cases, to escape.
|
| Figure 5. Highlighting a bottleneck in the evacuation with orange lines. This corresponds to the bottleneck in one of the hallways prior to evacuation. The bottleneck resulted in a lot of people attempting to exit through the southwest door. Due to this bottleneck, there was a casualty (RFID 59) who attempted to escape using an alternative exit. We also highlight some of the casualties near the northern exit (grayed areas), which shows us that most of these people where far from that exit. |