Dr. Sebastian Thrun, Professor of Computer Science and Electrical Engineering at Stanford University where he directs the Stanford Artificial Intelligence Laboratory, went over the steps his team has made in developing a self-driving vehicle at RoboDev in Santa Clara today. He showed some incredible video of cars smashing into obstacles (sometimes even seeking other cars out to smash into) but ended with videos of their latest vehicle successfully navigating slowly around other moving cars.
The great thing about his presentation was his appeal not to the side that wants self-driving cars, but to a side we can all agree with — saving energy, lives, and time.
In saving energy, Dr. Thrun explained that 22% of the Nation’s energy consumption is used by cars. You also only use your car on average during about 10% of your day, making it useless the other 90%. If self-driving cars could be developed, one car could be used by multiple people. “You could be dropped off at work and then send the car back home to pick up your wife.” Added safety will also increase gas mileage since removing the extra weight of safety features (airbag, reinforced steel) would increase fuel efficiency by 30%. (It should also be noted that convoys reduce energy consumption by 11%-17%)
We’ve seen some amazing robots recently. There’s the robotic tuna fish that will hopefully revolutionize the submarine world, there’s the super-realistic cod developed in Japan which still creeps me out, and of course let’s not forget the giant robotic spider that made Liverpool it’s home until it was herded into a tunnel by flamethrowers, hopefully never to be seen again (that thing still gives me nightmares).
The idea that these robots could be used by the military is very realistic. And while robotic fish are a great choice (imagine thousands of silent torpedoes, swimming around the ocean, looking for enemy ships), a giant spider might not be such a great choice. It’s an easy target, doesn’t hide very well, and despite the terror of facing one, you could outrun it easily.
So what things in the world should the military imitate in their desire for the perfect robotic weapon?
Children:My personal favorite. The idea that a simple child could be a deadly robot just makes so much sense to me. I mean, why would you think that five year old huddled in the corner in fear is actually programmed to rip your throat out?
Hornets: Already feared by all, the technology involved in making a hornet capable of delivering a poison sting, or possibly performing recon on enemy sites is too great to pass up. You could let a million of them loose on the countryside, spanning entire continents, looking for any sign of enemy activity (or even spying on other countries in peacetime).
Bats: It’s been tried before in World War II with live bats strapped to bombs (it didn’t work, go figure), but robotic bats would be stealthy and unnoticed. Their primary use would be night surveillance since any other creature flying around at night would be incredibly suspicious. On top of that, they could roost during the day, recharging their batteries with the Sun.
Snakes: Snakes are stealthy, can move efficiently on the ground, and have incredible senses. Now this could mean you could use it for surveillance, crossing a mine field,even silently taking out guards. Don’t forget there are sea snakes too. The only problem you’d run into is if you tried to invade Ireland, whoops.
The video you see here is of a robot made by MobileRobots.com using the MobileRanger Stereo Vision System. “MobileRanger stereovision systems are top-of-the-line instruments for measuring depth for demanding applications such as mobile robot navigation, people tracking, gesture recognition, targeting, 3D surface visualization and advanced human computer interaction.” You can see how objects at different ranges are represented by different colors (see my hand?). Very cool.
Above you see a photo from the display Boston Engineering had. What you see is a robotic fish they hope to build in the near future (sorry, no prototypes yet). I’m going to stay in contact with these guys on the project since it’s a pretty cool concept that could be built fairly quickly with the latest technology (the fact that they’re basing it off a Tuna fish is proof alone that this thing will be fast and powerful).
Robo Dev 2008 has kicked off with a free breakfast (yay!) and not free Wi-Fi (son of a…). Yep, in the heart of Silicon Valley and I had to shell out $13 to get internet for the day. But I’m not bitter (I’ll slash some tires on the way home today).
The first speaker of the day is Tandy Trower, General Manager of the Microsoft Robotics Group. He started off by saying “It’s hard for me to answer what exactly a robot is.” His presentation is about collaborating with parters in order to develop software to run robots on.
He showcased the Microsoft Robotics Studio which is a downloadable program that lets people design and program virtual robots. “Virtual robots are programmed the same way physical robots are.” His hope is that designers and programmers will build robots on their computers which can then be built in the real world.
By the way, although Tandy said he didn’t expect to see robots like the one above for another ten years, he says that you can expect to see it on the market next year for around $3,000. Anyone have spare pocket change?
Interesting Robot Fact
1. It’s wheels have other little wheels around it, allowing tires to move in any direction regardless of position.
2. It runs on Windows XP and dances to Britney Spears (ha).
For those interested, Mielle Sullivan and I will be live-blogging the Robo Development Conference and Expo tomorrow (Tuesday) in Santa Clara. Check the site for latest breaking robotic news as well as photos and vids from the conference. Also check out our Twitter updates under usernames jheylin and mielle_s.