Talking Traffic Episode 23 – Autonomous Vehicles
Hello and welcome to another edition of talking traffic. My name is Bill Ruhsam and I am your host. I operate this podcast and its sister website, talkingtraffic.org. Today is Monday, August 4, 2008.
In today’s episode, I’m going to ramble. Usually, Talking Traffic is about specific information that can help you or inform you *right now*, but not for this edition. Today, we’re talking about the future of highways and vehicles. Today, we’re talking about the inevitable revolution where our robot overlords will care for us crazy humans. How will we achieve this blissful state of utopia where all of our needs will be supplied, whether we like it or not? Well, it starts with a government contest.
In 2002, the Defense Advanced Research Projects Agency, otherwise known as DARPA, announced a competition that would take place in the Mojave Desert in 2004. The goal of the competition was to develop an vehicle system which could traverse a 200 mile course autonomously through the desert. By autonomously, they meant entirely by itself, with no human input whatsoever during the run. The vehicle was required to pilot itself and navigate the course, avoiding obstacles from start to finish. The course included both road and off-road terrain. No vehicle made it more than 8 miles before getting stuck.
In 2005, they did it all over again, this time with 5 vehicles successfully completing the entire course. Yay for them! Why was there such a big difference between 2004 and 2005? Well, I watched a program on the Discovery Channel back in 2004 which talked about how the various teams were designing their vehicles in order to overcome some of the challenges. Let’s just say that producing a vehicle that can analyze its surroundings and react in an appropriate manner is not for the faint of heart. Or maybe, I should say that riding in a vehicle that is making its own decisions is not something I’ll be doing any time soon. KITT, the artificially intelligent Ford Mustang which is star of the telemovie Knight Rider, is nowhere near realization in the real world. It took the teams two years to develop their Grand Challenge vehicles to a point where they could just barely leave the start line. Another year brought them all the way to the finish, but it wasn’t exactly fast. There’s a lot of processing that the computers on board those autonomous vehicles are required to do.
Let’s think about what is required for a person in order to drive from their front door to the supermarket. We’ll ignore quote unquote housekeeping like fueling the vehicle, turning it on, etc, and concentrate only on driving. You, the driver, must correctly place yourself in your surroundings, which requires both sensors (eyes, ears) and cognitive abilities to recognize what it is you’re sensing. It doesn’t do you any good to see a flat paved surface unless you recognize that the surface is (say) a road as opposed to (say) a basketball court. Then you have to know where you are with respect to where you’re going. I suppose it might be a lot easier to just drive down roads without any inclination as to a direction, randomly turning when you are at an intersection, but you’re trying to get to the supermarket right? And you’re starting from your home, yes? So you have to know where you are. Then you have to know traffic laws and be able to follow them. None of these items are easy for a computer to do, without assistance, and that doesn’t even get into the issues concerning the hardware required to do the sensing and the driving. I don’t want to get into all the niggly details on this podcast, but if you’re interested, try googling “autonomous vehicle”. You’ll have plenty of reading, I promise.
How might this affect us in the future? There have been several proposals and projects, again I won’t get into the details, see wikipedia, but some of these have been interesting. For example, there has been a proposal to use onboard radar to allow vehicles to be closer to each other than they currently get on high speed highways. If you close up the gap between vehicles, more vehicles can fit!
Currently the measure of congestion on a limited access roadway is density, and that’s measured in number of vehicles per lane per mile. There are lots of studies that show as density goes up, congestion does too, but what if you could increase the density without increasing the congestion? What if instead of approximately 150 feet between vehicles you could reduce that to 50? You could increase the capacity of a road by 3 without building anything! You’d be the Times person of the year!
The way this would work is to have a car automatically sense the speed and acceleration of the vehicle in front of it and close up the gap to something that would be unacceptable if a human were driving. Unfortunately, the technical challenges are legion, so this hasn’t happened yet.
That is just one example of an automated driving system. There are more. A good question, though, is Why? Why would we want autonomous vehicles? One reason is what I just mentioned: Congestion reduction. Another, personal reason, would be to reduce the amount of time that I, personally spend driving. I’d much rather be surfing the web, or reading a book than driving. If I could do that during my morning and evening commutes I would, damn betcha! Safety is also a good reason. As the saying goes in my line of work there is no such thing as an accident; someone is always at fault. If you could reduce the number of human errors that occur, you’ll reduce the number of collisions therefore increase the safety of everyone. The only way anyone has found to reliably reduce the number of human errors is training, or removing the human entirely from the decision loop. Autonomous vehicles would allow that. Of course, there’s always the human programming the vehicle who could screw up, or someone could maliciously invade the software, but I’m ignoring those scenarios right now.
How about cost savings? Wear and tear on vehicles would go down, as would maintenance needs on the roads. Increased safety means fewer collisions which means lower insurance rates and lower indirect societal costs. Fewer collisions also means more police performing law enforcement duties rather than responding to wrecks. Autonomous vehicles would probably be less inclined to drive like jackrabbits which would reduce emissions and raise air quality. There are all sorts of reasons why we would want robots to control our vehicles; these are only a few.
To sum up, autonomous vehicles would have a raft of benefits both personal and societal but unfortunately the technology has a long way to go.
This podcast is distributed under a creative commons 3.0 attribution non commercial no derivatives license, so you can give it away to the people you like but don’t change it or sell it and please remember to credit me and Talking Traffic.org. If you have any feedback or questions, you can email me at Bill at Talkingtraffic.org, or you can leave a comment on the show notes at talkingtraffic.org.
The music you hear is by Five Star Fall and can be found at Magnatune.com. Thanks for listening and we’ll see you next time. Have a great week.