I had a funny experience on my way home from work last week. I was on i280 and spotted a Google self-driving car ahead of me in the adjacent lane. As I drove up and then past it I noticed that it was maintaining a safe distance from the cars in front of it and falling farther and farther behind the passing cars.
Anyone who drives in moderately to heavy traffic commute traffic knows that there is a distance you can maintain behind the cars in front of you that is not by-the-book safe but prevents the phenomena where the gap is large enough to allow car after car to slip in front of you… it’s kind of like a traffic bullwhip effect. You drive just far enough behind that you can panic stop based on what is going on in front of the car you are following but not far enough back that you keep getting passed.
This particular Google self-driving car was, apparently, not coded with that rule and the result was rather comical. As a new car slipped in front of the self-driving car it dutifully dropped back to create a safe space which then became a new space for a different driver to slip in to, and so on and so on, the result being that the Google car dropped back at a predictable rate while car after car whisked by.
This is the challenge for all next generation technology to overcome, which is the requirement to adapt to situations that develop based on activities and patterns that are emerging in realtime. It’s not just a matter of more sensors and faster reaction times but a fundamentally different way of looking at software frameworks, and truth be told I have no insight to what the Google car is built on but one thing is clear, being able to parallel park or get from point A to B without incident is the least of their challenges.
Lastly, I am really excited about the prospect of self-driving vehicles. As much as I enjoy driving there is no doubt I would equally appreciate flipping into self-driving mode so I can take a call or read something or simply check out on my way home. On the commercial side, self-driving vehicle technology can remake logistics networks and shift commercial traffic patterns to have less impact on commute periods or reroute dynamically based on events that are happening. It’s exciting stuff, I think Google deserves credit for launching this experiment but the major auto manufacturers should also be recognized because they have been working on this longer than Google, despite getting far less attention.
(Cross-posted @ Venture Chronicles)