photo by Uber
We knew this day would come. The Tesla autopilot incident a few years back was a harbinger, but since a human was still “in control” it really wasn’t the test that autonomous vehicles had coming. But the AZ incident is the first case of a pedestrian being killed by a robocar. The AZ governor has been quick to say Uber isn’t to blame. As he should – since actually he is the most culpable in the whole incident. But there are three clear failure points in this beyond blaming the victim for jaywalking.
Fail 1 – Uber’s hardware and software. This incident wasn’t an “edge condition” where algorithms often are brittle and fail. Rather this situation of a human crossing the street at night outside of a crosswalk should be in the “known cases” for the system. While I am not an expert in autonomous cars, I do understand sensors and algorithms. Let’s start with the sensors. The military has been using thermal sensors to locate humans for how long? Seems like an oversight not to have that as part of the sensor package. As for the algorithms, reports are the car was doing 38 in a 35 zone and didn’t brake at all. Something certainly wasn’t hitting on all cylinders.
Fail 2 – the human in the loop. Another weak point in the chain that should be anticipated. Felony convictions aside, given the incredibly high percentage of humans who are distracted while driving, why should we expect a “co-pilot” to be any different? At the core of this is the normalization of phone addiction and the lax enforcement of distracted driving. We have slowly moved towards more draconian enforcement and penalties for drunk driving, why should distracted driving be any different? As someone who commutes by motorcycle daily, and due to sitting higher, able to see into people’s cars, I quickly lose count of the number of drivers who are fiddling with their phone rather than paying attention. Really – you’re not that important.
Fail 3 – the AZ government. Technology is seriously outstripping policy, but the AZ governor actually courted Uber and other companies after they were stymied by CA. Turns out the CA politicians wanted the companies to jump through hoops to test autonomous cars in the wild. AZ’s governor by contrast, prioritized income over safety. This clearly isn’t the first time a politician has used that calculus, and I’m not arguing against the advance of tech. Clearly autonomous cars are our future, and I believe they will in fact be safer than human drivers – particularly distracted human drivers. But the development of the technology needs to take place in a thoughtful manner, and not be driven by private industry or politicians looking for easy fiscal relief.