Regulating the Safety of Taxicabs and Uber and Lyft Cars by Algorithm

A few months ago I plugged a device called “Automatic” into the diagnostics port hidden under my car’s steering wheel, and it’s been collecting data about my driving habits and the health of my car’s engine ever since. Automatic reports this data to a website that presents it in a such way that I can understand. Automatic can also use this data to trigger simple algorithms in response to certain driving-related events. Playing around with this device and its capabilities gave me a glimpse into what the future of safety regulation for taxicabs and Uber and Lyft vehicles might look like.

This is what the Automatic device looks like. Some would call it a “dongle.”

Automatic and its competitors (with names like Dash, Mojio and Zubie) offer a basic form of “telematics,” the umbrella term for all the wireless communication between a car’s computer brain and the outside world. GPS navigation is a form of telematics, as is that feature of OnStar where they know if you’ve been in a collision. Imagine if your car had its own Twitter account, but it just constantly tweets what it’s doing and how it feels, several times a second. That’s telematics.

The data collected and reported by Automatic is fairly comprehensive. Automatic keeps track of my car’s location. It maps the routes that I drive (providing hard evidence that I almost never, in fact, go west of Vermont). It also records my driving behavior, and puts a little ding on the map where I accelerate too fast, or speed, or brake too hard. (Apparently the stop signs on Griffith Park Boulevard keep catching me off guard.) Automatic measures my fuel economy, adds up how much I spend on gas, and knows when I’m low on gas. It also knows if I’m ever involved in a collision. Automatic knows if my car’s “check engine” light goes on, and can tell me the exact nature of the problem if this happens.

A screengrab of my Automatic dashboard. Proof that I never go to the West Side.

It is easy to set up Automatic so that these events trigger an automated response. The Automatic companion app reminds me to fuel up and will text an emergency contact if I crash into something. Automatic also syncs up nicely with the web service, IFTTT, which allows me to program simple “recipes” that use the data reported by the device as a trigger for an algorithmic response. For example, there’s a recipe for “if check engine light is on, then email my mechanic and request a maintenance appointment.” This recipe is far more useful than the one I set up, which alerts members of a Slack room every time I park at Guisados for lunch.

Automatic told this Slack room when I drove to Watts to check out Locol. It’s very tasty!

This process, by which an inspector looks for problems and triggers a consequence if it finds a problem, is exactly how we regulate taxicab safety. Our method is far more analog, of course. A mechanic, police officer, or Los Angeles Department of Transportation official periodically checks the brakes, steering, headlights and “check engine” light of a taxicab, among other things. If something doesn’t work the way it’s supposed to, the taxicab is taken out of service. There it remains until the problem gets fixed.

The taxicab inspections conducted and required by the City of Los Angeles are infrequent and expensive. LADOT inspects each taxicab once per year, with occasional surprise inspections. We require our taxi companies to conduct their own weekly inspections, but these are of the “just eyeball it” variety, and we don’t strictly enforce this requirement, because doing so would cost an extra $3 million per year (according to one estimate I received). Should anything go wrong with a cab in between inspections, neither the City nor the taxi company would know about it.

The LADOT inspection checklist.

A telematics system like Automatic, by comparison, inspects a vehicle several times per second, whenever the car is running. Of the 34 things the City checks during a taxicab inspection, 7 of the most important can be monitored by telematics. If this data were reported to the City, we would always know if a vehicle was operating safely. Moreover, telematics can reduce the cost of inspections and maintenance borne by Los Angeles taxicab companies. When the package delivery company UPS added telematics capabilities to its fleet of trucks, it learned that preventative maintenance was unnecessary in many situations. By waiting to replace a part until after the telematics identified a problem, UPS saved millions of dollars a year.

Uber and Lyft’s vehicle inspection forms

The City of Los Angeles does not have the authority to conduct safety inspections of Uber and Lyft vehicles (the State of California kept this power for itself), but even if we did, we don’t have the resources to do it the same way we do for taxis. Of the 19 things that Uber and Lyft check for in a vehicle inspection, telematics can monitor 6. It’s estimated that there are more than 40,000 Uber and Lyft vehicles on the streets of Los Angeles, compared to less than 4,000 taxicabs. Adoption of a telematics system would solve this scale problem since we’d rely on a computer to monitor the safety of these cars, rather than human inspectors. If anything ever went wrong, say, if a “check engine” light lit up, the telematics system could trigger an algorithm that simply cut off a driver’s access to the Uber or Lyft app until the problem was fixed.

All this could happen automatically.