Driving Your Algorithm To Work

B1405-LH148

Recall the movie, “I, Robot.” Recall the sleek, automated cars and freeways, no-hands driving, parking which tucks away vehicles safely, and robots which assist humans to supposedly be better…well, humans. Now recall Google’s Honda, the self-driving car which the giant company is busy driving all over hell’s half acre with no human behind the wheel. Not gonna happen in MY lifetime, you say? It’s way too science-fiction-y to every become a serious way to get to your work (assuming you have a job to go to). But wait: now comes Ford, that iconic carmaker. It announced a partnership with MIT and Stanford to develop self-driving, automated, intuitive cars. Intuitive, as in a form of self-consciousness which rivals a wife’s suspicions when her husband goes to the gym and starts losing weight, as when a child knows which of his parents to ply for that toy she wants.

In his case, this intuition will come with some impressive bells, whistles, and infrared lights. This system (called “LiDar”) will bounce infrared light off objects as far away as two hundred feet and then—get this—will draw a real-time 3-D map of the vicinity, but that’s only the start. These sensors are powerful enough and sensitive enough to be able to discern the difference between a paper bag and “a small animal.” No word yet from the Furry Little Critter Protection Society about fine-tuning to make the car swerve away from gerbils. Ford wants their cars do what drivers do when they are behind a large vehicle which blocks the view ahead: maneuver back and forth inside the lane, trying to see around.

That 3-D pic will certainly help, although on Los Angeles freeways, the 3-D will look like one massive block of unmoving stone, but no matter—it’s the tech that counts. If the car in front of you suddenly slams on its brakes, your vehicle would “look around” and “decide” whether it’s safe to change lanes to avoid slamming into the back of that idiot in front of you. In other words, when this all comes to fruition, you will drive in a virtual space AND an actual space, being alerted to the reality around you, using virtual reality as a guide—a surreal concept, for sure. An article in the Los Angeles Times says this research with MIT “employs algorithms that help the car predict where moving vehicles and pedestrians will be.” Presumably a crystal ball wasn’t sexy enough to do the predicting, and Ford couldn’t find enough Gypsy fortune tellers to act as passengers for its prospective drivers.

Ford’s current vehicles can self-park (those of us old enough to remember our first driving tests, the one to get that coveted license, will remember with varying shades of horror and frustration, learning how to park between two cars on a busy street. Thank you, Ford), self-drive in slow-moving traffic (isn’t that what we already do?), and redirect drivers around heavy traffic (no word yet on whether Ford has a cure for the drivers merging onto freeways at sixty miles an hour while texting their friends that the party was s-o-o rad). The final step Ford will take is to fully automate their vehicles. OK, I’m no Luddite. I fully appreciate not having to go down to the well to lug a couple of buckets of water up to the cabin; I fully appreciate having a satellite dish bring me a football game; I fully appreciate writing this on a computer. What I don’t fully appreciate is NOT driving my car.

I don’t want a 3-D view of my driving world. I want a real-time view of it. I didn’t spend a half-century of learning myself and my limitations and strengths so I could let a computer decide whether to go slow or fast, whether to go around the weaving drunk in front of me, or whether I should sit back and read the sports page while my algorithm decides what to do. There have more than a few problems with airline pilots who have had their automated controls go haywire and didn’t quite know what to do. It’s called “forgetting how to fly” by some airline pilot veterans. They worry that the technology of airplanes is taking that very intuition and experience Ford claims is in need of phasing out way from pilots and putting the lives of passengers and crews at risk.

Of course, Ford says it intends, with other car manufacturers, to bring self-driving vehicles to the road by 2020. I wonder if this is such a good thing: recently, the upper Midwest experienced one of the worst spells of winter in recorded history. Will algorithms prevent skidding on black ice? Will any 3-D view of a whiteout be helpful, more helpful than some melodious voice from the car saying gently, “Pull over, moron, I can’t see a damn thing!” And what about those roads out there? This is not the world Will smith drove in. It’s a world of bumpy, fractured, pot-holed, narrow, poorly-lined, roads, filled with people driving beaters they bought in a vacant lot from a guy wearing shades. It’s filled with people who are struggling to find work and will not be able to afford a car which is so high-tech, it needs to come with a hacker. There are no sensors in these roads; there are no ways to keep someone from having too much to drink (or smoke) from getting out on the highway; there are people living in places with NO pavement and NO street signs of any kind.

When I first moved into my rural home, a county sheriff drove into my yard and asked me what street (actually, a dirt road) I lived on. Later in the week, I got a book from Amazon, and I asked the UPS driver if he had trouble finding my house. “No problem,” he said. So, I told my wife, if we have a crime incident, don’t call the cops, call the UPS guy. Maybe ford should use THEIR algorithm instead of MIT’s. Until the U.S. modernizes its infrastructure and gets serious about automated roads and cars, until it gets serious about automating its drivers, I’ll just climb into my old truck and drive myself.

Return to Bylines