The “black box” is often regarded as the Holy Grail of evidence when investigating a crash. But is the black box always right? The following case study demonstrates the importance of examining black box data with an underlying knowledge of its limitations, and always considering the data within the context of all other available evidence.

Let’s consider the following case: On a cold and snowy winter evening, police and emergency responders arrived at the scene of an accident where two individuals had lost their lives, two vehicles (and an ejected occupant) were in snowbanks, and another vehicle that was wrecked beyond recognition was still on the road. The initial investigation suggested that a large black SUV rear-ended a small red sedan (Figure 1A); the sedan was pushed into opposing traffic and was then struck by an oncoming white coupe (Figure 1B). The SUV and coupe both came to rest in the snowbank on their respective side of the road (Figure 1C). Tragically, the driver of the sedan and the driver of the coupe did not survive the wreckage, so the only people who could relay what happened just prior to and during the crash were the occupants of the SUV.

case study was the black box wrong-01

Figure 1A: A large black SUV rear-ends a small red sedan.

 

case study was the black box wrong-02

Figure 1B: The red sedan goes into oncoming traffic and is struck by a white coupe.

 

case study was the black box wrong-03

Figure 1C: The black SUV and the white coupe each go into the snowbank on their side of the road.

The occupants of the SUV suggested that the red sedan was stopped in the middle of the road, when the SUV’s driver came up on it, braked to attempt to stop, but slid on the snow covered surface and struck it.

As the involved parties wrestled over liability, the impact speed of the SUV and sedan became critical – but witnesses’ stories did not match up neatly. Go figure! Experts were brought into litigation on the case to analyze the evidence, which included SDM (Sensing Diagnostic Module or “black box”) data from the involved SUV. The SUV’s black box recorded the following information:

  • The SUV’s speed 5 seconds before the event was 89 km/h
  • The SUV’s speed 1 second before the event was 39 km/h
  • The speed change (a measure of impact severity) experienced by the SUV was recorded as only -0.02 km/h
  • The SUV was braking during the 5 seconds before the recorded event

The black box data was relied upon by an opposing expert to uphold a sequence of events that was very different than what the initial police investigation depicted. The expert argued that the sedan must have crossed over the centre line and collided with the oncoming coupe first, was pushed back into its lane, and then was struck by the SUV from behind. This opposing theory could, theoretically, make sense when looking only at the black box data. But what happens when you examine the scene evidence, the vehicles’ rest positions, and their crush damage?

We conducted a complete “traditional” collision reconstruction using the principles of the conservation of momentum, energy, and computer simulations. Our analysis determined the SUV’s impact speed when it struck the sedan to be between 104 and 114 km/h. We also estimated SUV’s speed change (a measure of collision severity) from the collision with the sedan to have been 15 to 25 km/h based on the damage exhibited by both vehicles. But as noted above, the black box recorded a pre-impact speed of 39 km/h and a speed change of only -0.02 km/h. So which is wrong — our analysis or the black box data? This is a trick question. And the answer is: neither is wrong.

You may have seen a version of the information systems hierarchy shown to the right. We’ll start at the bottom of this pyramid in our search for the reason there might be incongruity with the black box data and our reconstruction results.

Barring an electrical issue in the vehicle, the data wouldn’t be wrong. It’s just data – a bunch of numbers. The data from the SUV looked something like this:

$01 A0 52 00 000000
$02 C0 5A 00 000000
$03 41 53 37 30 32 39
$04 4B 36 42 52 32 31
$05 02 41 4F 4B 4A 00
$06 15 25 87 65 00 00

Our crash data retrieval tool interprets the encoded data stored by the black box into categorized information, which must be expertly examined. There is little chance of an error in this interpreted information (the tool has occasionally misinterpreted the category of data, but that is usually discovered quickly and fixed with the next software release). It was our expert knowledge of the data limitations and capabilities of this vehicle’s black box (based on engineering research and testing) that allowed us to rely on the information correctly. The experience (or wisdom) of an accident reconstructionist will assist him/her in noticing inconsistencies in the information and explaining them, especially as they utilize all the available evidence and compare the results.

So how can we explain why the impact speed we came up with was more than double the speed recorded by the black box? The SUV’s black box data was found to be unrelated to the collision with the sedan; instead, it was related to the SUV’s collision into the snowbank. How do we know this? The black box of the involved SUV could only hold one non-deployment event (recorded when the vehicle senses higher than normal accelerations, but not high enough to command airbag deployment). Since that particular black box only had one data “slot,” it would have overwritten any information in that slot should there be another non-deployment event (even if the severity is less). In this case, the event where the SUV struck the sedan was likely recorded to the black box, but then that data was overwritten by the SUV’s less severe collision with the snowbank.

Lending even more certainty to our conclusion that the black box data did not correspond to the SUV’s collision with the sedan, was the fact that we could use the black box’s recorded speeds for the 5 seconds to mathematically calculate the distance the vehicle travelled, which added up to exactly the distance between impact (with the sedan) and final rest position (in the snowbank) measured by the police at the scene. Therefore, the speed of 89 km/h at 5 seconds before the “event” represented the speed of the SUV just after impact with the sedan (the collision with the sedan would have abruptly slowed down the SUV’s speed from the 104 to 114 km/h at impact to about 89 km/h immediately after impact.).

It would be convenient if the black box held all the answers, but it is still just one piece of the puzzle. Relying solely on black box data and neglecting to analyze how it fits in with all other available evidence could easily result in a gross misunderstanding of what actually happened during a crash. Expertise on how the black box is designed by engineers, along with a strong understanding of its capabilities and limitations based on real-world research and testing, allowed us to realize that the black box wasn’t wrong, it was simply recording data about an event other than the one of (most) interest. Nevertheless, the information that it did contain helped us defend our opinions in court and was still an integral part of our reconstruction!

References: 

Eade v. Brown, Ontario Superior Court of Justice File No.: 05-CV-282960CM2.

Share this article:

Share This
Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now

×