Ady Leverette was a designer and a principal at Fat Pencil Studio between 2011 and 2018.
A few weeks ago, TriMet and its bus manufacturer New Flyer settled a major lawsuit involving "the worst tragedy in Trimet's history" for $4 million. The case stemmed from a high-profile April 2010 crash in downtown Portland. Five pedestrians were struck by a Trimet bus, killing two young women. The remaining three survived with moderate to severe injuries.
Fat Pencil Studio was hired by plaintiffs' lawyers Kirklin Thompson & Pope LLP, and Hala J. Gores, PC to reconstruct the path of the bus as it made a left turn across a lane of traffic into the crosswalk where it struck the pedestrians. We did this using a photogrammetric process involving footage from the bus security cameras and a detailed digital 3d model of the intersection. These findings allowed us to create a real-time SceneViz demonstration of the bus driver's view, which was significantly obstructed by the left side mirror and the "A-pillar" portion of the bus frame frame.
An example of the photogrammetric process that helped us establish the path of the bus.
This project was one of the most intense and comprehensive we have ever undertaken. It required us to dig deep into our toolbox to study the problem from every angle and achieve the greatest degree of accuracy in our findings. Our work to determine the path of the bus was aided by police investigation findings including: (1) a Total Station survey which showed the final position of the bus in relation to the intersection, and (2) the installed height of the forward facing on-board security camera.
Given a single known location, we used SketchUp's Advanced Camera Tools to reverse engineer additional details (focal length, tilt, roll) about the camera configuration. This allowed us to establish accurate camera positions at other times by matching reference points (such as building corners and light poles) seen in the video with geometry in the 3d model. We did this for 18 time-stamped frames of video footage during a 14 second period immediately before and after the collision, enough to describe the bus path and velocity profile in some detail.
A diagram showing the bus velocity in average miles per hour between photo-matched positions:
The lead pedestrian of the group was visible in several frames of the bus camera footage. The remaining four were walking close behind in a formation that was later documented in a filmed reenactment. This gave us enough information to accurately place all five pedestrians in the 3d model, at times corresponding to the video stills. At this point we had everything we needed to produce a 2d top-view animation. We chose this style of presentation because it accurately portrays the incident without getting into details such as body movements that would prove distracting for people trying to grasp the sequence of events.
You can see an excerpt from this animation below:
With the bus path reconstruction complete, we spent our remaining time on the case investigating what the driver could have seen or not seen immediately before the collision. We did an extensive analysis of forensic photos and data and took our own measurements to construct a detailed 3d model of the bus cockpit and driver. After incorporating this into our 3d model of the intersection, we could position a camera at the driver's eye height, and "look around". SketchUp's ability to move the camera in real time allowed us explore how driver head movement might affect the visibility of pedestrians at any given time.
Forensic measurements and bus path reconstruction allowed us to create these images from the driver's perspective in the seconds leading up to the crash:
This analysis was especially useful in understanding how the rear view mirrors and the "A-pillar" of the bus frame created significant blind spots for the driver. By tracing the line of sight from the driver eye position, we were able to draw an "obstructed view" area and compare it with the progress of the pedestrians in the crosswalk. The results confirmed what many had guessed after viewing the security camera footage: the pedestrians remained in the driver's blind spot for the entire left turn maneuver.
The A-pillar (top left) and side-view mirror (top right) significantly obstructed the driver's view throughout the turn:
This case settled the night before trial was scheduled to begin, so our work was never presented in court. It was however extremely useful during trial preparation. "The animation was hugely convincing," said attorney Steve Thompson. "Focus group members sat stunned as they watched the impact occur in the crosswalk. We heard audible gasps from several of them."
The 3d model visibility study was also valuable to other experts retained on the case. "Understanding what the driver could see in the seconds before the crash was crucial," said Dr. Gary Sloan, a human factors expert. "This ground-breaking work made it possible for me to offer more substantial testimony in the case."
One final takeaway from our experience working on this case: a different mirror configuration would have made it a lot easier to see people in the crosswalk. The cost of equipment and re-training for switching TriMet buses to one of these alternatives would be a significant one-time expense. However, this seems trivial when compared with the human toll exacted by left turn blind spot collisions.
A driver's view of the crosswalk with three different mirror configurations: high mount (left); existing (center); inverted mount (right):
Understanding what the driver could see in the seconds before the crash was crucial. This ground-breaking work made it possible for me to offer more substantial testimony in the case.
Dr. Gary Sloan