Re: Building a Robo-Magellan bot
- I don't have any creds yet at Robo-Magellan but have been building one for this year's competition. I am using the 4 wheel version of the Minds-i vehicle. It has 4 wheel drive and 4 wheel steering so there is plenty of low end torque and the turning radius is tighter than you can get with an RC truck. The microcontroller is an Arduino mega 2560 with a Mega shield. I have five Ping sonar sensors on the front and one on the back. There is of course a GPS sensor and a compass. I also added a Memsic tiltometer to keep the vehicle from going into a steep ditch or too steep a hill. Instead of a camera system that requires a lot of weight and fancy programming I am using a color sensor that simply reports wether the object is orange or not. If it is orange, then there is a bumper on the front of the vehicle that will gently touch the cone without knocking it over or moving it. The robot now also knows to go to the next GPS site. It is a more simplistic approach than others are taking but the kiss principle has always worked well for me. Of course there are a lot of complicated behaviours I have to program into the system but it would take a lot of space here to give you a complete idea of what I am doing.
--- In SeattleRobotics@yahoogroups.com, Tony Mactutis <tony@...> wrote:
> From what I saw last year, all you need to be in the running is a
> mobile platform that can drive on grass and sidewalks and follow GPS
> The bar may have been lowered a bit from previous years but I think that
> will be a good thing, if it encourages new people to try.
> On 2/18/2013 6:56 AM, Steve Hassenplug wrote:
> > Bob,
> > The biggest mistake I see people make, is that they try to build the
> > perfect robot on their first attempt.
> > If you just start with a cheap 4wd RC car/truck, you can design a
> > system that can simply control two servos (one for steering, another
> > for the speed controller). That will allow you to have a moving
> > robot, without much trouble. Down the road, you can upgrade the
> > chassis, if you wish.
> > For sensors, I would recommend (in order)
> > GPS
> > Compass
> > Camera (cone detection)
> > Object sensor(s) (ultra-sonic)
> > Wheel Encoder
> > Touch/bump sensor(s)
> > The controller is really wide-open. If you Google these types of
> > robots, the main thing you'll find is that they all use /different/
> > controllers. My current favorite controller is my cell phone, and a
> > connected IOIO board for I/O.
> > In '07 I brought a robot to the Robomagellan in Seattle. It was about
> > 1 month old, and didn't do very well. I came home, and convinced
> > Chibots to run the event. In '08 Kennith was there as my robot
> > completed the first Chibots' course.
> > This year, that same robot got third place in the event you read
> > about, while my newest creation (still only a few months old) got a
> > respectable fourth place.
> > My advice: Start now. Get something rolling. Then, you have more
> > time to figure out "why is it doing that?"
> > Good luck
> > Steve
> > On Sun, Feb 17, 2013 at 12:50 PM, Bob Cook <bob@...
> > <mailto:bob@...>> wrote:
> > Hey everyone,
> > The Robothon event is still seven months away and I've been
> > wanting to build an entry for the Robo-Magellan event ever since
> > the first year it was run. I figure there must be others out there
> > with the same desire and maybe already making great progress.
> > There was a great article in the recent Robot magazine issue about
> > Chibots' event that made for interesting reading.
> > Here are some thoughts I've had about how to make an effective robot:
> > - Start with an RC truck, preserving as much of the built-in
> > hardware and electronics as possible, to save time building up the
> > chassis
> > - Expect to drive over long grass as well as hard surfaces, so
> > bigger is better
> > - Will need to avoid obstacles like trees and people, so some sort
> > of ultrasonic rangefinders are will be helpful
> > - Physical bump sensors are probably nice to have but don't seem
> > strictly required
> > - GPS won't be accurate enough as the only navigational aid, but
> > could provide useful information
> > - ArduPilot might be a useful platform to build upon (although I
> > have yet to look at it, might be too specific for planes)
> > - Finding the cone seems to be a challenge best solved visually
> > e.g. a web cam + OpenCV
> > - Some sort of remote "kill switch" is going to be required
> > - Testing, testing, testing and more testing
> > I've seen the very impressive works of David Anderson (jBot) and
> > Mark Curry (Intrepid and Nomad) and others as well.
> > I'm really curious to hear what other people are doing e.g. what
> > do you think of my approach?
> > Are you are building something for Robo-Magellan?
> > - Bob