After hearing about the DARPA competition and success of many of the robots, I’ve decided that I would like to build a smaller one of my own. I want to start off small and just have it be able to travel about 100 feet, so it will most likely be an RC car sized robot, which means no giant sensors hanging from the bumper. I was wondering if anyone knew what sensors the teams used, and where they can be purchased. Hopefully I am not going in over my head, and if you have any advice, I’d love to hear it!
If you’re interested in what sensor’s different teams used, look at www.grandchallenge.org, which is the competition’s website, or at the individual teams websites. I remember looking briefly last year at the wide variety of devices used. They ranged from gps systems to complex (I assume) laser scanning systems and radar systems. I don’t know what you’re capable of, but some of this stuff would be well over my head I know.
That said, if you’re interested, supposedly you can get GPS to work with a simple microcontroller. I know this was covered at least briefly in one issue of Servo Magazine if that helps. Besides GPS, there are sensors that won’t break the bank and are small enough to fit on a r/c car that will accomplish the same goals. Infrared and ultrasonic sensors both can serve to tell you if there’s anything in front of the car, and accelerometers, gyroscopes, and digital compasses can provide some idea of how your robot is oriented and are available online at many robotics sites. I would suggest looking at www.parallax.com.
Good luck, building an autonomous robot is an interesting problem, nad has applications that would serve a FIRST team well.
Most teams used the following:
Differential GPS: gps with a land based correction signal for accuracy of inches instead of 10-15 ft (like the garmen ones you can buy)
~$3,000 + cost of correction service
LADAR: Laser distance and range finder. A single beam laser that spins along a spectrum at 180 degrees to map terrain in front of vehicle. (most teams used ones made by SICK)
~$6,000
Stereo optic camera set up: allows the computer to sense distance just like your eyes (uses at least 2 cameras)
~$500 per camera
Other sensors used:
Encoders - ~$50 a piece
Radar - no idea we didn’t use one
Laser point sensors - no idea we didn’t use any
The darpa challenge was extremely expensive to compete in, Stanford had a $500,000 budget and CMU was close to $10,000,000. What I am trying to say is don’t get discouraged. There are other ways to do obstacle avoidance and stay on track. For your project I would say you are going to need a GPS and probably at least a sonar/radar system to work well.
Three words: FIRST Grand Challenge
Many teams also used inertial navigation for times when GPS went out. These can also be used for sensing tilt angle. I compete in a competition called IGVC (intelligent ground vehicle competition) which is a very close thing to darpa, but our robots have a minimum length of 3 feet and the course is only 10 feet wide at the widest parts. You may be able to find some info here http://www.igvc.org/deploy/ We use Sick LMS (laser measurement system) now, however in the past we used an array of 8 sonar units from Polaroid cameras.
BTW our darpa vehicle can be found here
http://team.cartlink.org/
I hope to post some news about igvc soon, also igvc was mentioned in a post by Jack Jones
http://www.chiefdelphi.com/forums/showthread.php?t=38640&highlight=igvc and yes I am proud to be the member he spoke of.
I can see ith now…Autonomous Segways must complete a 25 mile course through city streets during rush hour :ahh:
Check out Robo-Magellan. The robots are less than 50 pounds and must fit in a 4 foot cube. The straight line distance from start to finish is less than 300 feet, although it is not possible to follow a straight line due to obstacles. This sounds like what you’re interested in.
As I recall the original rules came from the Seattle Robotics Society . They ran a competition last weekend at Robothon in Seattle. The Portland Area Robotics Society also ran a competition at PDXBOT back in April. The PDXBOT site has a number of pictures. Several of the entrants look to be based on RC trucks. Several of the PARTS members participate in Robo-Magellan and there has been some Robo-Magellan traffic on the PARTS mailing list recently. I’m sure the folks there would be happy to offer advice.
Greg
if you have a mindstorms kit or two around, play around with sending the IR signals from the rcx and then use the light sensor (or multiple light sensor) to have an IR “radar” thingy… i havent really worked too hard on this as im not much of a programmer but i have tried it enough to prove that it does work. obviously it works better on whit walls and not so well on skinny obstacles… but if you worked hard its definitly a way to have a lego robot capable of avoiding obstacles without having to come in contact with anything. makes for faster running…
I always thought it would be fun to make a robot that can autonomously traverse the W&OD bicycle trail.
- Traffic is fairly slow and attentive (compared to me
) - It has a variety of slopes (enought to be challenging)
- Problems like crossing the street and avoiding cyclists/joggers would prove… interesting
A great site for assorted electronics bits is Marlin P. Jones & Associates. I buy a bunch of things I can’t find at radioshack (Like N-connectors for wifi projects) from them.
I can think of a few things you can do, I will also b trying these as soon as I get my rc car built from the chainsaw motor sitting on my garage floor and the 24 feet of steel tubing in my basement.
–Use an IR or RF beacon in stead of GPS if you are on a tight budget.
–Parallax has a article on their website about serially interfacing a $100 GPS with a BASIC Stamp
–Accelerometers and gyros to autonomously navigate terrain, along with distance sensors.