23 thoughts on “NOVA’s The Great Robot Race

  1. I think it’s cool stanley won. If the documentary is correct, then it followed my prefered robotics paradigm. Brains over brawn.

    Their method of looking at the terrain coming up using a camera is cool. It matches the close laser scans to pixels on the camera shot, giving it an idea of what the road looks like, allowing it to figure out where the rest of the road is, in the distance. The red team had a complicated hardware long-distance laser system.

    I don’t know what the preoccupation with lasers is. A laser/camera combo as in stanley is ideal.

  2. “greygoose:umm why is this on hack-a-day?”

    its not a ‘hack’.. its an update for us (not counting you) like minded people how are into this sorta thing. The technology in this challenge is amazing, something we’d all love to be able to do ourselves some day, or at least can respect those who are able to do it.
    If you want to bitch about the content on hackaday, make something and send it in, it’ll get posted if its worth a damn.
    complaining about free information is like wiping your ass before you shit… just makes no sense.

    watch the show before you comment, you’ll understand why eliot decided it was worth its own “extra” post.

  3. My best friend’s father was the one of the main engineers on the CMU (Carnegie Mellon University) team. It sucks that they lost, there was a glitch in the gas tank when it switched to the resevoir tank from the main and it wasn’t getting all of the gas it needed. That one was supposed to be the speedy one, and of course since it couldnt get gas on the uphills it was basically dead. The other was set at only a set speed. That is why it only gained a bit on Stanley’s. All in all im proud of all the teams for the work the contributed, i think this is one of the coolest things ever!

  4. bah, we haven’t even come close to inventing everything… There are areas for new ideas all around, new technologies coming out every day that allow for even more new ideas. If you can’t see opportunities for new things, then you should re-think about how you look at the world.

  5. the grand challenge makes me giggle with excitement. it’s awesome factor is through the roof. anyways has darpa said anything about another grand challenge? the mission was “accomplished” this year so they said there wouldn’t be another. i dunno, the only reason vehicles finished this year was because there were double the number of GPS waypoints as the first year. maybe i’ll find out once i’m done torrenting the show. oh, and props to nova for being an excellent program.

  6. I watched it as well. Its a bit unorthidox for Hackaday, but I appreciate it, i think there is few of us that dont like this. Read this story in wired a few months back though.

  7. Stanley deserved to win. They simply had a better design. No fancy gimballed laser to break, simpler hardware design, less expensive but quite capable vehicle, and most of all:

    They didn’t have a team of 30 people “cheating” by adding thousands of additional waypoints between the supplied waypoints. Their vehicle actually navigated the road betweeen each supplied waypoint automatically.

  8. how do you think cmu ‘cheated’? You dont think that the military would do satellite images and determine even better routes around a coordinate set that they were provided. Before you assume that they cheated maybe you should take on this task.

  9. Technically cmu didn’t cheat, since adding waypoints wasn’t against the rules. That’s why I put the word “cheated” in quotes.

    But the point of this competition was to make a vehicle that could follow roads and go around obsticles. CMU’s vehicles had to do much less work to figure the route since they added thousands of waypoints. This was demonstrated when their laser unit got stuck turned to the side and the vehicle still finished the race. They had so many waypoints they didn’t need to do any scanning of the road to see where to drive. That wasn’t the point of this competition.

    And no, I don’t think the military would use satellite images to “determine better routes round a coordinate set” unless that was the only way this could be done. Stanley demonstrated that this wasn’t necessary. Ideally you want to be able to just give it a set of directions from a map and let it go.

    I also get really tired of people who say things like “you assume that they cheated maybe you should take on this task,” and other forms of the standard “if you can’t do better then shut up.” There is nothing in the electronics on these vehicles that I couldn’t have designed, but I’ll admit that my software programming skills probably aren’t up to the task. But just because I couldn’t do what they did doesn’t mean I don’t have the right to to criticize what they did.

    Now, since most of this sounds rather negative, I would like to say that this race was a great accomplishment by everyone who was involved, and all the entries were quite impressive. The military, and anyone else interested in autonomous driving, will learn a great amount of valuable knowledge about what to do and what not do do when designing systems of this type, which was the ulitmate goal of this exercise. Congratulations to everyone involved.

  10. Yeah, cmu approached it the wrong way. I’m glad stanley won, it approached the problem in the spirit of the competition.

    Yeah, the adding in of tons of waypoints was cheating in my opinion (not by the rules, just by the spirit of the competition). It probably took 60 man-hours to create all of those additional waypoints. It would take a human less than 5-hours to remote control it around the course.

    What’s the point of autonomy if in practice it takes a ton more effort to do anything?

    This is why I think the military preoccupation with autonomous robots is retarded. Sure, it saves lives to have robots do things. Why not make them remote control? still saves lives, and gives all of the good gamers of the world decent jobs.

    Yes, this would introduce strategies such as jamming the control signals, in which case autonomy might be necessary, however most of the time you could probably use remote control.

    @13 – this is why cmu cheated – their robot doesn’t really solve a problem. It would be easier just to remote control it, instead of devoting 60 hours of work creating waypoints.

  11. you realize that these waypoint werent the only thing used. it had several sensors, even when the gimbal was stuck sideways the sensors were still guiding the car, but it was low range only. The waypoints gives the car a general way to go, which is the same thing you do when you go from one place to another by car, you get directions. Considering the way the system works, these “thousands of waypoints” were not used to steer the vehicle, just to give it an idea of where to go. If the waypoints said one thing, and that path had a huge obstacle around it that wasn’t planned for, it’d entirely go around that obstacle and get back on track, which was specified by the waypoints.

  12. hi my name is mathew jones and i am a mechanical engeneer for team axion racing. unfortunately we did not get much footage to nova even though we qualified to start off fourth at the darpa GC 2005. our vehicle is the 94 jeep cherokee with the two surfboards on top and the spinner hubcaps.

    cmu “red team” also had two vehicles in this race which all in all were still led by whittiker. they did add waypoints last year and this year as well as move them to theyre maps.

    the point of this race was to be given a solid set of waypoints, upload them into your vehicle and push “go”.

    not to take the information…get 30 people to edit it and add waypoints and at the last minute upload the waypoints…..they even had a vehicle out in the desert for a long time just yo premap every god damn road they could find….

    red team had well over 3 million doallars to use on this project whereas my team had a little over 400 thousand. we are also using many more types of sensors than they are so our obsticle detection is well more advanced wich is greatly going to help us in the next challenge..The Urban grand challenge will consist of moving obsticles possibly stop signals and more.. it is believed that we are already set for this next race.

    they might have completed it with theyr millions…stanford may have won…but we have the real intellegence in our sensor fusion!!

  13. wow, probably 60 hours of programming more waypoints all done within the 2 hours of time they had! I want their time machine slower downer device :P

    Part of the DARPA race states that the route is unknown until 2 hours before the individual start times.

    So it might not be 100% autonomous (thats up for debate) – but it does have one very nice feature… redundancy… If I were to be the person buying this machine for use in the military. I want to know that when something goes wrong (guaranteed) that the machine will get the task done.

  14. “wow, probably 60 hours of programming more waypoints all done within the 2 hours of time they had! I want their time machine slower downer device :P”

    Dude – they had 30 some people in that room splitting the route into “segments” and working on it for 2 hours. That’s what he meant by 60-man hours, how can you not see a flaw in that plan? What kinda got to me is that they didn’t just add the waypoints, they adjusted the speed of the vehicle ahead of time meter-by-meter for the whole route. So they pretty much did remote control it!

  15. This post is a bit late, so hopefully ppl are still following.

    I would just like to say the race was a milestone in robotics. Im sad the event hasnt received much attention from the media. I seriously compare these achievements to Man attempting to fly.
    Im happy Stanley won because I wanted brains to beat brawn. And I think I agree with the whole

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.