Since access to automotive “black box” data first became available in the early 1990s, they have increasingly become a gold mine of information for insurers and investigators alike. What we colloquially refer to as a “black box” in an automobile is officially known as an event data recorder (EDR) (sometimes also called the airbag control module –ACM; or sensing diagnostic module – SDM). “Black box” is a borrowed nickname from flight data recorders on airplanes. However, these systems work quite differently; while flight data recorders save all kinds of data from the entire trip made by the plane, automotive EDRs typically record their data in a continuous loop of temporary memory, writing over the information every five seconds or so until an “event” occurs (we’ll get to what those “events” are below). The reason EDRs record data in a loop is because they weren’t put inside of vehicles to record every movement that happens in transit the way aviation flight recorders do. Rather, the primary function of a vehicle “black box” is actually to sense forces and accelerations, and then to decide, within milliseconds, whether or not to deploy the airbags. Lucky for us, the data surrounding an airbag deployment decision is usually incredible helpful to investigators and reconstructionists (as long as it makes sense with the other available physical evidence).  Moreover, these modules have evolved from a simple diagnostic and energy reserve devices to a data-rich component with increasing utility for forensic crash reconstruction.

What information do EDRs contain?

The only time an EDR records anything to its internal memory is if higher than normal everyday forces or accelerations happen. There are two primary types of “events” that EDRs record: deployment events and non-deployment events.

When the EDR senses higher than normal accelerations and decides to deploy the airbags, it will record what we call a “deployment event.” This event, along with a snapshot of some data like speed and brake or throttle use from a few seconds immediately before the airbags “went off,” is locked in the black box and cannot be overwritten. Anytime airbags have deployed in a vehicle, the chances are extremely high that its black box will have at least some information about what happened right before those airbags deployed.

Black boxes also record what are called “non-deployment” events when the vehicle experiences higher than normal accelerations that were sufficient enough to “wake up” the system, but when it ultimately decided against airbag deployment. Examples of when we commonly see non-deployment events are more minor parking lot fender-benders.

Like deployment events, non-deployment events on a black box also save some data from a few seconds right before the “event.” But what’s crucial to know about what makes non-deployment events different is that this information is much more vulnerable and can be overwritten. There are generally two situations that could result in a non-deployment event vanishing from a black box: some other “event” of either type, or after a certain number key cycles. There are a few exceptions to these rules, but the general takeaway remains that non-deployment events are extremely vulnerable to being cleared if they are not swiftly preserved after the incident in question.

Do all vehicles have “black boxes”?

General Motors pioneered the way with the introduction of accessible EDRs in the early 1990s. Most modern vehicles manufactured in the 2000’s contain some sort of black box, but the vehicle manufacturer has to have allowed for black box information access in order for us to obtain EDR data. Thankfully, the last few years have seen the vast majority of auto makers offer support for EDR imaging. Nevertheless, auto manufacturers each developed their own safety restraint systems, and as such their EDRs use their own proprietary algorithms and record different parameters. As claims will involve vehicles of different makes, models, and ages, there is a wide variety (read: inconsistency) to the specific information recorded even when EDR data access is available, including the circumstances that result in an event being recorded. That can even mean that a crash of the same severity could result in an airbag deployment event on one vehicle’s EDR, but not on another’s. It is for these reasons that the type of information available from an EDR download varies depending on the year make and model.  For a current list of what vehicles support black box downloads, go to

What kind of information can we find on an EDR when an event has been recorded?

Generally, these are the most common types of information we can get off of modern vehicles when we “image” or “download” the EDR: speed, speed change, brake and throttle use, and seat belt use. Newer vehicles can record even more specific information along with an event, such as individual wheel speeds and tire pressures, odometer reading, gear position, and steering input.

“Imaging” the EDR involves retrieving a copy of the electronic information stored on it, and when done by a properly trained EDR Operator, the data on the black box itself should not be cleared or spoiled in any way. 

The Black Boxes of the Future

Semi- and fully autonomous vehicles are already being tested and operated on our roads and setting off tectonic shifts in the transportation, insurance, and legal landscapes. The promise of safer roadways with fewer accidents is a noble one, however, the transitionary era will still see many crashes. Even in a completely driverless world, collisions will still occur as long as there is a steering wheel and a brake pedal (albeit at a much lower rate). As automobiles are becoming increasingly autonomous, they require more embedded sensors that read and store more information than we have ever seen before, including (but not limited to): camera images, driver profiles, and distances from nearby vehicles, roadway markings, traffic signs/ lights, people, and other objects. The proliferation of data that will be recorded to semi- and fully autonomous vehicles’ EDRs (or some future manifestation of them) will no doubt be crucial in understanding how people are injured and property damaged.

Currently, more than 90% of car crashes are the result of human error. Testing with semi and fully-autonomous vehicles is demonstrating the enormous potential for giant leaps in roadway safety. For example, the Insurance Institute for Highway Safety (IIHS) reported that vehicles equipped with automatic braking systems reduced rear-end crashes by about 40 percent on average! As sophisticated machines take over more and more driving tasks, the question won’t be which driver is to blame?, but rather, is the manufacturer to blame? Product liability cases will become more prominent as investigations into autonomous vehicle glitches and decisions come under scrutiny. As a result, EDRs will play a crucial role in untangling fault in those disputes (provided that all of the EDR data is openly accessible; since consistent data access has been a problem with EDRs in the past and present, this challenge will most likely continue into the future).

In the meantime, even in vehicles without auto-pilot or automatic emergency-braking capabilities, the general trend has been to record more and more data, and in the future, we expect EDRs to have more information, capture data for a longer amount of time leading up to an event, and record more events.

Regardless of the data available, it must be forensically analyzed in the context of the physical evidence in order to make sense of the information. In our on-going crash testing research, we have instrumented vehicles in real world crash scenarios to compare their imaged EDR data with other instrumentation so that we can pinpoint the reliability, validity, and accuracy of the EDR data in different scenarios.  Doing so allows us to expertly examine real world crashes by making sense of what the EDR data, in conjunction with all other available physical evidence, is trying to tell us.


While “black boxes” may seem like the holy grail that will give you all of the answers you need to an incident in question, things are a bit more complicated than that, and they will remain that way until many more data-rich autonomous vehicles are on the roadways. Although black box data is very reliable in most scenarios, and can even be the “smoking gun” in an investigation, they have limitations and there are circumstances where EDR data can be misinterpreted (for example, the event on the EDR may be related to a collision other than the one in question). That is why black box data always needs to be forensically examined and compared with all physical evidence to make sure it is being properly contextualized. When used correctly, black box data allows for reconstructions with greater detail and higher certainty. Still, they are just one tool in a reconstructionist’s tool box. Most crash reconstructionist experts know this, but the people who depend on them should know it too.

Share this article:

Share This
Your browser is out-of-date!

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