Google Robocars at TED, Shanghai bubble cars, Robot Week, lives saved


Here's a few Robocar updates.

First of all, the TED talk given by Sebastian Thrun, leader of the Google self-driving car team (disclaimer: they are a consulting client) is up on the TED web site. This is one of the short TED talks, so he does not get to go into a lot of depth, but notable is one of the first public showings of video of the Google car in action on ordinary city streets. (The first was at PodCarCity, but video was not made available on the web.)

At TED the team also set up a demonstration course on the roof of a parking lot, and allowed some attendees to ride and shoot videos, many of which are up on the web. While the car does perform well zooming a slalom course, and people have a lot of fun, the real accomplishment is what you see video during the talk.

Another "City of the future" video has appeared featuring robocars prominently. This Shanghai 2030 video plays out a number of interesting robocar aspects, though their immense elevated road network reminds me more of retro futurism. A few things I think will be different:

  • The people in the car sit side-by-side. I think face-to-face is much more useful. It's more pleasant for conversation, and it allows for a narrower car which has huge advantages in road footprint and drag. Some people can't stand facing backwards, and so there will still be side-by-side cars if you have two people like that, but I think a large fraction of cars will move to face-to-face, either narrow (for 2) or wide (for 3 or more.)
  • The video shows cool displays projected onto the windscreen. This "heads up" sort of display makes sense if you have to keep your eyes on the road while using the screen, but in these cars, the people don't. On the other hand it's true that some people get motion sick looking down while riding, but you can also put an opaque screen in the middle of the window in a robocar.

It's National Robotics Week with lots of robot related events. In the Bay Area on Thursday, an all-day robotics demo day for kids and adults will take place at Stanford's robotic car lab, so people will get a chance to see Junior and other Stanford robocars there.

Fatalities drop

The trend continues -- last year U.S. road fatalities dropped again to 32,788. That's a steady decline since over 43,000 5 years ago. And this is in spite of total vehicle miles going up. As a result, the death rate per 100 million miles is now 1.09, the lowest it has been in 60 years.

That's very good news, though many forces fight for the credit. The leading contender seems to simply be that cars are getting safer in crashes, with better crumple zones and air bags, and more people wearing seatbelts. Medicine has also gotten better. Some will also be coming from better cars with safety systems like anti-lock brakes, crash-warnings and lane-departure warnings -- precursors to robocar technology -- but it would be wrong to assume these are a big component. Also worth noting that this happens in spite of the rise of people talking and texting while driving, though the secretary gives some credit to the recent laws banning this. But that doesn't explain why the drop began in 2005.

It's also odd that while fatalities drop almost everywhere, they're actually up in New England by 18% and by 4% around the midwestern Great Lakes, and generally up around the north-east.


The link you have to the TED talk links to some thing called "yugma" that asks to install stuff on my computer. No way.

A better link is

That's the bane of having two clipboards on linux. I had the TED link in my main clipboard and the java conferencing tool I was trying out (and do not recommend, by the way) was in the other clipboard! I was fixing it as you commented.

Thanks to your link and "in spite of total vehicle miles going up" comment, I just realized that I won my challenge to this bet:

2010 did end up with *slightly* more miles driven than 2005, but the trend in miles driven is no longer a consistent increase. It's been a bumpy road, so to speak.

Actually, your bet should have been pretty safe. If not for the economic collapse, I think you would have won more handily. While the oil price spikes have indeed been high, perhaps higher that was suggested, they just seem to drive car efficiency -- nobody wants those SUVs as much.

One interesting statistic is that in the last 25 years, internal combustion engines got 30% more efficient, while average passenger fleet mileage got slightly worse, because people used that gain to buy heavier cars (SUVs.) As soon as the gas price spikes, they jump on the chance to use the efficiency. And there is even more efficiency that's come, from hybrids and lighter cars and even PHEVs.

"because people used that gain to buy heavier cars (SUVs.)"

As much as I want to blame the SUVs, compare the weight (and MPG) of a Honda Accord or Civic from 1986 with a 2011. Even compare today's Civic with the 86 Accord, since that's more fair. It's shocking how heavy today's "small cars" are.

The fatality data suggests that what we've traded for that lower fleet mileage is increased safety.

A lot of the safety design is actually not so related to the weight. It's just better design. Admittedly give them more metal to play with and they can do better, but it's more about designing how it will bend and crumple.

"The trend continues — last year U.S. road fatalities dropped again to 32,788. "

Germany has about 1/3 the population of the US and no speed limit on about half of its highways. The 2010 death toll was the lowest in several decades, about 2000 IIRC. This means that, adjusted for population, the U.S. death toll is about 5 times what it should be. OK, in the U.S. people drive more due to lack of public transportation, but on the other hand the density of traffic is probably lower on average than in Germany. What is the problem?

The death rate is actually much higher on city streets -- about twice as high per mile, and 4-5 times per hour, than on highways. So the speed limit issue on the highways is not the big factor. The rates vary wildly around the world. Check out the rate in the UAE, for example. (There, liability is based on seniority. If a senior official drives off the road and hits a pedestrian of lower status, the pedestrian has to clean pay to clean his blood off the car.)

Yes, I agree, most accidents don't happen on the highways. (Also in Germany, despite (or, perhaps, because of, since it means cars are built to handle it) the lack of a speed limit, most accidents and deaths are not on the highways. Still, even with public transportation, the typical European city has about the same amount of traffic than the typical American city.

Add new comment