Information continues to emerge about the automated vehicle in Uber’s fleet that fatally struck a pedestrian the night of 19 March 2018 in Tempe, Ariz. While Tempe police chief Sylvia Moir cautiously speculated that neither a human nor an automated driver could have avoided the crash based on video of the incident, she refused to rule out charges for the backup driver.
This was a bit puzzling—how can a driver be at fault in an unavoidable crash? Then, Tempe police released video of the crash on Wednesday night.
In the video, taken from the car’s forward-looking dash cam, the pedestrian seems to appear from nowhere wearing all black, on a dark street, outside of a crosswalk—a seemingly impossible situation for a driver. Meanwhile, the camera recording the inside of the vehicle showed the test driver looking down at something near the console for more than 5 continuous seconds, immediately before impact. Traveling at 40 miles per hour, the vehicle would have covered the length of a football field in that time.
Warning: This video contains content that may be graphic or disturbing.
Tempe Police Vehicular Crimes Unit is actively investigating— Tempe Police (@TempePolice) March 21, 2018
the details of this incident that occurred on March 18th. We will provide updated information regarding the investigation once it is available. pic.twitter.com/2dVP72TziQ
Moir acknowledged that her analysis was “preliminary,” and she’s correct. The National Transportation Safety Board is already conducting its own investigation that will look at sensor data and decision algorithms used by the automated driving system. If the data is accurate (Uber has been known to intentionally hide data in response to warrants), we should have a good understanding of what exactly went wrong in the next few months. Until then, any analysis is speculative.
That hasn’t stopped othersfromspeculating. Although the video shows the pedestrian appearing almost out of thin air, she would have, in fact, crossed two turn lanes, a through lane, and half of the Uber car’s lane before being struck—that’s roughly 42 feet. Walking at a speed of 3.5 feet per second, the design walking speed for traffic-light “walk” signals, she would have been on the road for more than 10 seconds before impact.
Even at a pace of 7 minutes per mile, fast enough to qualify for the Boston Marathon, she would have been crossing the street for more than 3 seconds. The lidar system used on Uber’s automated vehicle to detect pedestrians works just as well in complete darkness as in broad daylight, yet initial reports said that neither the automated driving system nor the test driver applied the brakes until after impact. So, what happened?
There are three ways the lidar might have failed: It never detected the pedestrian, it detected her but misclassified her as a benign object such as a bush or an exhaust cloud, or it classified her correctly but failed to accurately predict her movements, perhaps assuming she would stop in the turn lanes. Most experts seem to suspect misclassification, as someone pushing a bicycle with bags creates an unfamiliar point cloud for an automated driving system.
Without more data or a detailed crash investigation, any conclusions are also speculative. Even the information we have now may not be accurate—while police and most media outlets reported that the speed limit was 35 miles per hour, a Google Street View image from July 2017 clearly shows a 45 mph speed limit sign a mile before the crash site.
Several production vehicles offer optional driver fatigue and alertness monitoring systems. Eye tracking systems are inexpensive and widely available. Were these installed on Uber’s vehicles? Were drivers regularly monitored and their performance assessed? Monitoring driver attention is particularly important in highly automated vehicles, as studies consistently show that off-road glances increase dramatically when the throttle, brake, and steering are simultaneously under computer control.
Make no mistake, this was tragic, as is each of the 30,000 fatal crashes each year in the United States. Unlike so many accidents that are caused by human carelessness, this crash has the potential to improve the safety of advanced vehicles—if the automated-vehicle industry can learn from and improve on their failures. Uber must be fully transparent and candid with investigators and regulators about not only what happened but also what the company could have done better.
About the Author
Noah J. Goodall is a senior research scientist at the Virginia Transportation Research Council, in Charlottesville, Va.
Noah J. Goodall is a research scientist at the Virginia Transportation Research Council, in Charlottesville, Va.