Opinion: what a self-driving car can teach us about... using indicators

One of the profoundest gifts of human consciousness is the ability to introspect. By watching ourselves and our fellow humans we are able to learn and advance as a species. Yet I was recently taught something about human behaviour not by another human but by a machine. Admittedly that machine, complex though it was, had been programmed by humans.

I was being driven by an experimental autonomous Nissan. We were circumnavigating a roundabout and I asked the engineer whether this car, which obviously has visual technology of extreme acuity, can see other vehicles’ indicators and act accordingly. He snorted and said no, that’d be far too unreliable. Oh dear: for my whole driving life I’ve clung to the notion, even in the face of endless evidence to the contrary, that it’s not only worth indicating myself but also taking others’ indications seriously. This self-driving tech is telling me I’ve been wrong.

Nissan acts by the logic that indicating isn’t a necessary condition of a driver’s changing path. They can and do shift lanes or exit a roundabout without signalling. I guess we knew that, sadly. But the Nissan-bot has also been programmed on the basis that there can be false positives. You might see a vehicle indicating right and going left, or vice versa, or indicating but not changing course at all because the driver has knocked the stalk and has the radio on too loud to hear the click-clack.

Even so I want to use my indicators. They tell you I’m about to move into the lane you’re in, just in case you were about to accelerate and close the gap. I want to tell you I’m leaving the roundabout at the exit before the one you’re trying to join from, so you can join a little sooner – if everyone did that then cumulatively we could all get a useful amount of life back.

I want you to use yours, too. The Nissan programmer told me his robo-Leaf operates by “body language”. It sees where other vehicles are going, and sets its own path and speed to avoid them. But I still think that strategy makes for lumpier traffic flow. For instance cars with advanced cruise control are horribly lumpy. Imagine my reading of human body language were as bad as their reading of vehicles’ body language: if I walked through a crowded pub I’d end up doused in other people’s beer.

I drove a Mustang Mach-E with the BlueCruise hands-off function. We were sailing along at 70mph and a Mini pulled into our lane doing 50-ish. It indicated so if I’d been driving (not testing the outer limits of the system’s capability) I’d have seen its indicator and slowed down. The Mustang didn’t, so it did precisely nothing until at least a quarter of the Mini’s width had transgressed the lane I was in, and then it had to jam the brakes on. Or rather it didn’t because I, not trusting it, already had.

That’s why when I’m overtaking cars in a jam on my bicycle and they move without signalling, I smile at the driver and politely say, “excuse me, your indicator’s broken". My, how they’re grateful for that free advice.