Tech

Autonomous Delivery Robot Adds External Airbags for When It Inevitably Crashes Into People

From the dream of zero crashes to wrapping cars in giant pillows.
external airbag
Credit: Nuro
Screen Shot 2021-02-24 at 3
Moveable explores the future of transportation, infrastructure, energy, and cities.

Autonomous vehicle company Nuro, which makes car-like autonomous vehicles that drive on roads to deliver pizza and groceries mainly in suburban-like settings, has unveiled a new vehicle model that the company says it is ready to mass produce and “is approximately 20 percent smaller in width than average passenger cars” which “gives bicyclists and pedestrians more room to maneuver alongside the bot.” However, should the bot crash into said cyclist or pedestrian, it has an additional feature: a giant external airbag.

Advertisement

“A custom external pedestrian airbag across the front of the bot is optimized to reduce the force of impact and number of injuries in the event of collision,” Nuro’s Medium post on the new robot says. The next sentence says the vehicle’s top speed is 45 miles per hour, a “​​low speed,” according to Nuro, which “reduces the severity of any potential impact” but also “allowing the vehicle to reach more customers and serve more businesses on a wider variety of roads.”

First of all, 45 mph is not a “low speed” at which to get hit by a car. According to a 2011 AAA study, the risk of severe injury when hit by a car traveling at 42 mph is 50 percent, and at 50 mph the risk of severe injury is 75 percent. 

Nuro was founded by two former execs in Google’s self-driving-car unit. It currently has limited partnerships with Domino’s and Krogers to deliver in the Houston area. It also has approval from the California DMV to operate driverless delivery services in some Bay Area neighborhoods. Some competitors use much smaller robots that go slower but travel on sidewalks.

Advertisement

There’s good reason to be skeptical these airbags are little more than safety theater. The post provides no information whatsoever on how the airbag works, a critical bit of information considering the difficulty in engineering effective external airbags. 

Internal airbags save lives because they soften the so-called “second impact” when the car occupants smash against the steering wheel, dashboard, or windshield. To prevent this second impact, airbags inflate after sensors on the car detect a crash. In about 30 milliseconds or less, the airbags inflate and provide a cushion between the driver and the car’s hard internal surfaces. Airbags work because they have time to inflate between the first and second impact. 

But in order to have any effectiveness whatsoever, external airbags would have to inflate prior to the first impact. In other words, the bot has to predict the future. And if it can do that, perhaps it would be better to avoid hitting the person entirely. 

Many cars have crash detection technology nowadays that automatically provide a warning or apply the brakes when the car thinks it is about to crash into something. And this can, indeed, be life-saving technology. But as anyone who has ever used a car with this feature knows, it can also produce many false positives and false negatives. “Phantom braking” has become a phenomenon in modern cars due to this flawed technology. 

While an external airbag that works well in the event the robot crashes into a pedestrian or cyclist is theoretically better than no airbag at all, it also reinforces the expectation that the robots will crash into people and that these crashes are an inevitable part of autonomous vehicle operation, just like crashes are an inevitable part of driving a car. This is perhaps a more honest approach than we’re used to hearing from the AV industry, but it’s also a stark departure from the future it has been selling. Once upon a time, the autonomous vehicle industry was bragging about how their cars would be so safe they were going to eliminate car crashes entirely. Now that they’re resorting to wrapping their cars in pillows, that seems like an awful long time ago.