Team Jefferson and the Self-Driving Scion
07-26-2007, 03:31 PM
#1
Junior Member
Thread Starter
Join Date: Jul 2007
Location: Crozet, VA
Posts: 2
Team Jefferson and the Self-Driving Scion
What's the craziest modification you've considered doing to your Scion? Full-kitchen with running water and gas stove? Installing a hot tub? Replacing all of the body panels with flat-screen televisions and cameras to make it turn invisible? (That last one's probably a bad idea, if it worked)
I'm a project manager for Perrone Robotics and we, along with some students from the University of Virginia, have decided to modify our Scion xB in a way that is maybe not as crazy as the ideas above, but maybe is a little crazier.
We want our Scion to drive itself around a city.
Our car is named Tommy Jr., and he's performing in the DARPA Urban Challenge. The goal of the UC is to take upwards of 80 teams and see how many of them can get their vehicles to the point where they can drive themselves in a city, with other moving vehicles, obeying traffic laws, without, you know, hitting anything. The course is 60 miles long, and we have 10 hours to complete it.
Now, not all 80 teams will make it. It's been whittled down once to 53 teams, and we've recently shown off our car to visiting DARPA people to show how well Tommy is doing so far. (DARPA is the Defense Advanced Research Projects Agency, a branch of the Pentagon for trying to do, basically, science-fiction things). On August 10, we'll learn if we've made it to the National Qualifying Event which takes place in late October. Roughly 30 teams will make it there, and probably around 10 will make it to the final event in early November.
Why am I telling you all of this? Same reasons everone else posts to the board. We think it's cool, and want to show off. We want to inspire other people to try to think of new ways to make their Scion xBs better. And because, from time to time, we run into problems, and when that happens, we like to ask you for help.
The great thing about Team Jefferson is that, unlike other teams, we don't have millions of dollars to spend on getting a robot working. So far, we've spent about $50-60k, including the car and paying people to work on it. All of the parts are commercially available. Everything is within reach of a few clever people who want to spend a little more money than average on a very custom car. Oh, sure, it's a bit more expensive than the general hobbyist project, but I've seen others that cost more over time.
Anyways, this was just a quick (ahem) introduction. Soon we'll post some more details about the actual ways we've taken Tommy apart and put him back together again. Better. Maybe stronger. And probably quite a bit slower. But definitely better.
Brian J. Geiger
Project Manager
Team Jefferson
Perrone Robotics
I'm a project manager for Perrone Robotics and we, along with some students from the University of Virginia, have decided to modify our Scion xB in a way that is maybe not as crazy as the ideas above, but maybe is a little crazier.
We want our Scion to drive itself around a city.
Our car is named Tommy Jr., and he's performing in the DARPA Urban Challenge. The goal of the UC is to take upwards of 80 teams and see how many of them can get their vehicles to the point where they can drive themselves in a city, with other moving vehicles, obeying traffic laws, without, you know, hitting anything. The course is 60 miles long, and we have 10 hours to complete it.
Now, not all 80 teams will make it. It's been whittled down once to 53 teams, and we've recently shown off our car to visiting DARPA people to show how well Tommy is doing so far. (DARPA is the Defense Advanced Research Projects Agency, a branch of the Pentagon for trying to do, basically, science-fiction things). On August 10, we'll learn if we've made it to the National Qualifying Event which takes place in late October. Roughly 30 teams will make it there, and probably around 10 will make it to the final event in early November.
Why am I telling you all of this? Same reasons everone else posts to the board. We think it's cool, and want to show off. We want to inspire other people to try to think of new ways to make their Scion xBs better. And because, from time to time, we run into problems, and when that happens, we like to ask you for help.
The great thing about Team Jefferson is that, unlike other teams, we don't have millions of dollars to spend on getting a robot working. So far, we've spent about $50-60k, including the car and paying people to work on it. All of the parts are commercially available. Everything is within reach of a few clever people who want to spend a little more money than average on a very custom car. Oh, sure, it's a bit more expensive than the general hobbyist project, but I've seen others that cost more over time.
Anyways, this was just a quick (ahem) introduction. Soon we'll post some more details about the actual ways we've taken Tommy apart and put him back together again. Better. Maybe stronger. And probably quite a bit slower. But definitely better.
Brian J. Geiger
Project Manager
Team Jefferson
Perrone Robotics
07-27-2007, 03:47 AM
#4
Senior Member
SL Member
Scion Evolution
Join Date: Oct 2006
Location: Pearl City, HI
Posts: 995
i drove a truck with a viper alarm on it. When you press the button on the remote it would say "viper armed" and "viper disarmed" and he even made his own " back off b****" i was like whoa! working valet is ok i guess. Lot's of mad people tho that just don't understand inflation and short handed on workers and stuff....
08-01-2007, 02:42 AM
#7
Junior Member
Join Date: Jun 2007
Posts: 23
Best of luck to Tommy Jr. and the whole team! I'm sure you guys are working hard, and I'm equally as sure you have the overall support of the entire Scion community.
Can't believe, however, that Toyota didn't kick you guys a free xB, or some sort of contribution. You're spearheading the kind of technology my kids will take for granted in 20 years.
Can't believe, however, that Toyota didn't kick you guys a free xB, or some sort of contribution. You're spearheading the kind of technology my kids will take for granted in 20 years.
08-01-2007, 05:59 PM
#9
Senior Member
Scikotics
SL Member
Join Date: Apr 2006
Location: West Virginia
Posts: 1,459
Nice job....this is defiently sometihng that i would like to see get updated
this would make an excellent driver-back-up system...driver falls asleep or w/e on the freeway, bam, secondary system could slow him down and safely off the road..
mad props
this would make an excellent driver-back-up system...driver falls asleep or w/e on the freeway, bam, secondary system could slow him down and safely off the road..
mad props
08-03-2007, 01:39 AM
#13
Member
Join Date: May 2007
Location: Charlottesville, VA
Posts: 38
Just found on the U.V.A. website:
http://www.virginia.edu/uvatoday/new...se.php?id=2455
Woot for the cutest AND smartest car on the planet!
http://www.virginia.edu/uvatoday/new...se.php?id=2455
Woot for the cutest AND smartest car on the planet!
08-03-2007, 03:37 PM
#14
Junior Member
Join Date: Jul 2007
Posts: 2
The sensors we're using are SICK lidars. They're basically a laser range finder that sweeps 180 degrees. When the beam hits something it gets reflected back, and we get the distance and the angle. From there we can calculate if the object is an obsticle or not.
Our audible alarms are quite simple, only going off when the car is driving by itself.
We tried to get Scion as a sponsor but, they didn't go for it. Not yet anyway.
And lastly we're running the stock steel rims with factory hubcaps. We got Tommy used and the previous owner could not parallel park if their life depended on it; the caps were badly curbed. So we painted them black with a silver rim as a temporary measure until we decide on what rims/tires we want for the race.
Our audible alarms are quite simple, only going off when the car is driving by itself.
We tried to get Scion as a sponsor but, they didn't go for it. Not yet anyway.
And lastly we're running the stock steel rims with factory hubcaps. We got Tommy used and the previous owner could not parallel park if their life depended on it; the caps were badly curbed. So we painted them black with a silver rim as a temporary measure until we decide on what rims/tires we want for the race.
08-03-2007, 05:46 PM
#15
Senior Member
SL Member
Scinergy
Join Date: Feb 2006
Location: GA
Posts: 614
I'll bet there are people on here near you who have rims they'd donate if you let them help/watch. Additionally, you may be able to subsidize some of the cost by offering the aforementioned "services" people are looking for (backup sensor, proximity sensor, etc.) I've heard of this contest before, and as I understand it, no one has ever completed the course as of yet. I'd love to hear about what type of "brain" you are running, and where you'd mounted it. I currently have a carPC, but I seriously doubt that a standard Mini-ITX MB with and embedded processor could handle the calculations you want. Great for GPS, music and video though 
08-07-2007, 09:26 PM
#16
Junior Member
Join Date: Jul 2007
Posts: 2
Actually in 2005 Standford won the second DARPA Grand Challenge with a VW Touareg, with Carngie Mellon coming in second and third with their two entries(H1 Hummers). Check out www.grandchallenge.org.
Tommy's brain resides in the back cargo area, and is surpizingly powered by a car pc with a couple of mirco controllers branching out to the actuators. It's all Java based and is run by an operating system we call MAX developed by Perrone Robotics. Think of MAX as windows for robotics. The base low level stuff is already taken care of (drivers for all the various and sundry sensor types). And what you do is install your application meaning, the specific code for your robot, and you're off and running. MAX is very configurable and adapts easily to new envoirnments. For example once we got the actuators and the computer hooked up; we downloaded the software from our First robot Tommy Sr. and within a couple of hours of fine tuning and calibrating, Tommy Jr. was born.
By the way I'm the teams mechanic, not a programmer. So you'll have to ask one of them for anything specific.
Tommy's brain resides in the back cargo area, and is surpizingly powered by a car pc with a couple of mirco controllers branching out to the actuators. It's all Java based and is run by an operating system we call MAX developed by Perrone Robotics. Think of MAX as windows for robotics. The base low level stuff is already taken care of (drivers for all the various and sundry sensor types). And what you do is install your application meaning, the specific code for your robot, and you're off and running. MAX is very configurable and adapts easily to new envoirnments. For example once we got the actuators and the computer hooked up; we downloaded the software from our First robot Tommy Sr. and within a couple of hours of fine tuning and calibrating, Tommy Jr. was born.
By the way I'm the teams mechanic, not a programmer. So you'll have to ask one of them for anything specific.
08-10-2007, 02:31 PM
#18
Member
Join Date: May 2007
Location: Charlottesville, VA
Posts: 38
08-10-2007, 04:08 PM
#19
Junior Member
Thread Starter
Join Date: Jul 2007
Location: Crozet, VA
Posts: 2
Technology in Team Jefferson's self-driving Scion
Today, I thought I'd tell everyone about some of the hardware that we're using in (and out) of Tommy Jr. But first...
We've made it to the semifinals! Wooooo!
Okay, that's out of the way. John, our "Satanic Mechanic", has told you a bit about what we have in Tommy. Of course, he would know, as he's put most of it in. Let's start with sensing. In order for the car to drive itself, it has to be able to see what's around it, know where it is, and know where it's going. As Tommy Jr. doesn't have eyes and an inner ear for balance, we have to make up for that in a variety of ways.
First, we have the LIDARS. The LIDARS are lasers that are infrared, so they won't blind anyone, that can tell us if something is in front of the laser within 200 meters. We have four of these, so Tommy can see all around the car. We'll know how wide the object is, how far away it is, and, by comparing from millisecond to millisecond, if it's getting closer or farther away. These are far more flexible than a proximity sensor, as mentioned in use in the Viper alarms, but also far more expensive. A proximity sensor will tell you if something is close, but it won't tell you where it is, how big it is, or if it's getting closer.
Next, we have a GPS. This is similar to the in-car GPS that many of us use to find our way from place to place, with three main differences: 1) the interface is wildly different, as it talks to a computer instead of a person; 2) it can tell us where we are to within a couple of inches, rather than several feet; and 3) it costs about 5-10x as much as a good GPS unit. However, having a good GPS makes up for Tommy Jr.'s inability to understand "where" he is, like we do. We can look around, read a street sign, see where the sun is in the sky, and a thousand other things to get a sense of where we are in the world. We may be terribly wrong ("We're not lost, we don't need to ask for directions..."), but we still think we know where we are. Tommy doesn't have that sense of place, so we make up for it with the very accurate GPS.
Third, we have a camera. It's actually two cameras together, what's called "stereoscopic". Having the two cameras is like having two eyes, and it gives Tommy some depth perception. Some of this will let him add to the information that the lasers give him, because some surfaces don't reflect lasers very well, and it would be a shame to hit something just because it was very black and dull. The other thing the camera does is lets us see lines painted on the road. Tommy Jr. still has to follow traffic laws, and we don't want him swaying between lanes just because the road is curvy.
Fourth is our Inertial Navigation (INav) system. The GPS is great for telling Tommy Jr. where he is, but you have to do some extra math in order to know which way he's pointing. The INav is like Tommy's inner ear - gives him a sense of balance and direction. Without it, he might not be dizzy, but he would be a little confused.
Our last sensor of note is the RADAR. RADARs are great because they will tell you where things are when they're far away, and sometimes we need to know if something is farther than 600 feet away. The downside of the RADAR is we can't see something if it isn't moving. This doesn't really hurt us, as if something is more than 200m away, but is standing still, it's unlikely to bother us. Still, it works differently that many of the other sensors, and how we work, but I suppose it's much like how a T-Rex sees the world (or so I'm told).
Next time, I'll tell you about the computers we use in Tommy Jr., and how you just might have enough power in your car already to make it autonomous.
We've made it to the semifinals! Wooooo!
Okay, that's out of the way. John, our "Satanic Mechanic", has told you a bit about what we have in Tommy. Of course, he would know, as he's put most of it in. Let's start with sensing. In order for the car to drive itself, it has to be able to see what's around it, know where it is, and know where it's going. As Tommy Jr. doesn't have eyes and an inner ear for balance, we have to make up for that in a variety of ways.
First, we have the LIDARS. The LIDARS are lasers that are infrared, so they won't blind anyone, that can tell us if something is in front of the laser within 200 meters. We have four of these, so Tommy can see all around the car. We'll know how wide the object is, how far away it is, and, by comparing from millisecond to millisecond, if it's getting closer or farther away. These are far more flexible than a proximity sensor, as mentioned in use in the Viper alarms, but also far more expensive. A proximity sensor will tell you if something is close, but it won't tell you where it is, how big it is, or if it's getting closer.
Next, we have a GPS. This is similar to the in-car GPS that many of us use to find our way from place to place, with three main differences: 1) the interface is wildly different, as it talks to a computer instead of a person; 2) it can tell us where we are to within a couple of inches, rather than several feet; and 3) it costs about 5-10x as much as a good GPS unit. However, having a good GPS makes up for Tommy Jr.'s inability to understand "where" he is, like we do. We can look around, read a street sign, see where the sun is in the sky, and a thousand other things to get a sense of where we are in the world. We may be terribly wrong ("We're not lost, we don't need to ask for directions..."), but we still think we know where we are. Tommy doesn't have that sense of place, so we make up for it with the very accurate GPS.
Third, we have a camera. It's actually two cameras together, what's called "stereoscopic". Having the two cameras is like having two eyes, and it gives Tommy some depth perception. Some of this will let him add to the information that the lasers give him, because some surfaces don't reflect lasers very well, and it would be a shame to hit something just because it was very black and dull. The other thing the camera does is lets us see lines painted on the road. Tommy Jr. still has to follow traffic laws, and we don't want him swaying between lanes just because the road is curvy.
Fourth is our Inertial Navigation (INav) system. The GPS is great for telling Tommy Jr. where he is, but you have to do some extra math in order to know which way he's pointing. The INav is like Tommy's inner ear - gives him a sense of balance and direction. Without it, he might not be dizzy, but he would be a little confused.
Our last sensor of note is the RADAR. RADARs are great because they will tell you where things are when they're far away, and sometimes we need to know if something is farther than 600 feet away. The downside of the RADAR is we can't see something if it isn't moving. This doesn't really hurt us, as if something is more than 200m away, but is standing still, it's unlikely to bother us. Still, it works differently that many of the other sensors, and how we work, but I suppose it's much like how a T-Rex sees the world (or so I'm told).
Next time, I'll tell you about the computers we use in Tommy Jr., and how you just might have enough power in your car already to make it autonomous.
08-10-2007, 04:23 PM
#20
Senior Member
SL Member
Scinergy
Join Date: Jan 2006
Location: Castle Rock, Colorado
Posts: 4,088
We have some guys using our parking lot to test out their Driverless car. It's really something to see.
Did great on the roads with other cars and such but the off ramps on the highway kept giving them problems due to it was just following the lines.
Did great on the roads with other cars and such but the off ramps on the highway kept giving them problems due to it was just following the lines.