ELEV-8 Quadcopter AeroMagellan Competition?
CowboyCoder
Posts: 33
So I had a brainstorming session with myself last night... that's never good. I came up with the idea of a RoboMagellan (SRS) type of competition just for quadcopters, an arial RoboMagellan if you will. I could hardly contain myself when I got to spring it on Ken and it was decided, in keeping with the open, community developed nature of the ELEV-8, that I should post the seed here in the forums and let you, the user, figure out what we should call it and a basic set of rules.
Below is a proposal only of two different classes, autonmous and R/C.
Keep in mind, the rules should be simple and firm for this to work. I am thinking we need to work towards something like one single printed page of rules?
Objective - Fly a course using GPS, locate waypoints, broadcast to the pilot that each waypoint found and return home. Obviously, we will want to identify and avoid obstacles in flight.
Maybe the R/C class could be a fly by video using a spotter just for safety?
Open to all quads, of any manufacture, that meet the weight limit rule?
First, the business end:
This needs a name. RoboMagellan is the property of SRS I think... oh,oh,oh I propose The Parallax AeroMagellan ? CMON, that's good and you know it lol.
I formally nominate Parallax as the 'keeper of the rules'. Everyone ok with that?
Now, the rules: (proposed autonmous class)
(Everything MUST fit with the current AMA rules)
1 - All quads must be under pilot control at all times for safety. This means that we can fly autonomous with a working radio in our hands that will allow us to stop, land or drive the aircraft into the dirt as needed to keep public and property unharmed?
2 - All events must be Line of Sight only?
3 - All aircraft must weigh less than 10 pounds?
4 - All flights must have a pilot and a spotter?
5 - Course length 1000 feet?
6 - Course will include a starting point, 5 waypoints and an ending point?
7 - Waypoint will be marked by some agreed upon color of some agreed upon dimension?
8 - Maybe institute the AMA R/C rulebook as it pertains to flying fields and locations?
9 - All aircraft must transmit live video to a central video device for public viewing and safety?
Now, the rules: (proposed R/C class)
(Everything MUST fit with the current AMA rules)
1 - All quads must be under pilot control at all times for safety. This means that we can fly autonomous with a working radio in our hands that will allow us to stop, land or drive the aircraft into the dirt as needed to keep public and property unharmed?
2 - All events must be Line of Sight only?
3 - All aircraft must weigh less than 10 pounds?
4 - All flights must have a pilot and a spotter?
5 - Course length 1000 feet?
6 - Course will include a starting point, 5 waypoints and an ending point?
7 - Waypoint will be marked by some agreed upon color of some agreed upon dimension?
8 - Maybe institute the AMA R/C rulebook as it pertains to flying fields and locations?
9 - 9 - All aircraft must transmit live video to a central video device for public viewing and safety?
OK guys and gals, it's all you now. Lets try to put together a common, fair rule sheet and then we can 'kick the tires and light the fires'. I got a quad and I am ready to compete!!
For the time being, this topic will serve as a repository for a final rule set. I will work to compile what ya'll agree upon into the final set and submit it to Ken for safe keeping.
Thanks Ken / Parallax.
3-2-1 GO!
Below is a proposal only of two different classes, autonmous and R/C.
Keep in mind, the rules should be simple and firm for this to work. I am thinking we need to work towards something like one single printed page of rules?
Objective - Fly a course using GPS, locate waypoints, broadcast to the pilot that each waypoint found and return home. Obviously, we will want to identify and avoid obstacles in flight.
Maybe the R/C class could be a fly by video using a spotter just for safety?
Open to all quads, of any manufacture, that meet the weight limit rule?
First, the business end:
This needs a name. RoboMagellan is the property of SRS I think... oh,oh,oh I propose The Parallax AeroMagellan ? CMON, that's good and you know it lol.
I formally nominate Parallax as the 'keeper of the rules'. Everyone ok with that?
Now, the rules: (proposed autonmous class)
(Everything MUST fit with the current AMA rules)
1 - All quads must be under pilot control at all times for safety. This means that we can fly autonomous with a working radio in our hands that will allow us to stop, land or drive the aircraft into the dirt as needed to keep public and property unharmed?
2 - All events must be Line of Sight only?
3 - All aircraft must weigh less than 10 pounds?
4 - All flights must have a pilot and a spotter?
5 - Course length 1000 feet?
6 - Course will include a starting point, 5 waypoints and an ending point?
7 - Waypoint will be marked by some agreed upon color of some agreed upon dimension?
8 - Maybe institute the AMA R/C rulebook as it pertains to flying fields and locations?
9 - All aircraft must transmit live video to a central video device for public viewing and safety?
Now, the rules: (proposed R/C class)
(Everything MUST fit with the current AMA rules)
1 - All quads must be under pilot control at all times for safety. This means that we can fly autonomous with a working radio in our hands that will allow us to stop, land or drive the aircraft into the dirt as needed to keep public and property unharmed?
2 - All events must be Line of Sight only?
3 - All aircraft must weigh less than 10 pounds?
4 - All flights must have a pilot and a spotter?
5 - Course length 1000 feet?
6 - Course will include a starting point, 5 waypoints and an ending point?
7 - Waypoint will be marked by some agreed upon color of some agreed upon dimension?
8 - Maybe institute the AMA R/C rulebook as it pertains to flying fields and locations?
9 - 9 - All aircraft must transmit live video to a central video device for public viewing and safety?
OK guys and gals, it's all you now. Lets try to put together a common, fair rule sheet and then we can 'kick the tires and light the fires'. I got a quad and I am ready to compete!!
For the time being, this topic will serve as a repository for a final rule set. I will work to compile what ya'll agree upon into the final set and submit it to Ken for safe keeping.
Thanks Ken / Parallax.
3-2-1 GO!
Comments
In this thread I had planned on doing some autonomous GPS naviation as a demo for the Expo, however I did not get the hardware in time to fully realize this goal and wasn't sure if enough people would be able to pull it off. I think it's a great idea, though detecting the 'cone' may be more difficult from an aerial perspective. I was going to stick directly to GPS coordinates. In fact, once done I plan on having mine deliver an S2 or other robot, which will then move on from the 'drop zone'. I thought this described a real-world use of this technology. I can't wait to see what others come up with.
Nice to meet you.
I think the cone could be replaced with something like we use for arial crop spraying. It is not uncommon to see crops marked with a cross laying on the ground, in white, so the pilot can find them. If we went with something, say 48" X 48" in orange, I think it would be a pretty easy find using off the shelf gear.
Too funny about delivering another robot, I just had a similar conversation last night about delivering a robot that would have to go 'touch' the 'cone' and return to the quad to be sent to the next location.
Lets brainstorm more here!
I had been putting my efforts into ground-based GPS navigation, much like the RoboMagellan, sans the orange cone detection. When the Quad-Copters started becoming popular I surmised that it may actually be easier to do this with an aerial robot, simply because being in the air, which having it's own set of issues, removes several issues that hinder ground-based GPS navigation, such as obstacles and metal objects, such as rebar under concrete which can affect the compass module.
Since most quad-copters have some mechanism for stable hovering, you could use that (modify) mechanism to allow autonmous navigation by programming the quad-copter to hover steady and level, and then changing what the it perceives as level. If you shift the bias of the leveling routine you cause motion with inherent stability. I was doing some experimentation with that concept, but didn't have practical hardware with which to experiment on. I think the theory is sound though and hope to try it this summer.
I think the middle point should also be a land, wait 15 seconds, then liftoff and resume. Points would be awarded for the accuracy of the landing (i.e. within 3 ft, within 6ft, within 10ft).
Need to make this interesting enough to garnish some real media attention.
Thinking aloud here... Does it need to be open to all micros or might it be restricted to Parallax the Propeller P8X32A as the main micro ?????
I'd also like to see various categories of contests that recognize creative or well engineered usage of the various processors and platforms that Parallax manufactures.
Of course, the proper winner of an AeroMagellan contest would be the one who perishes just before completing the voyage.
How about the winner is the craft that can leave the largest crater?
I think it should be open to all that wish to participate in order to foster a wider competition base?
I like the landing rule, great idea. What if it were exteded to all waypoints?
I like it a lot.
Chris,
This is where my thoughts are as well. Switching from hovering over a point to autonomous forward flight is easy yet difficult in my mind. First hover, thats built in, but then transition into forward flight which is going to cause a slight forard tilt. At this point the craft would want to drive that angle into the ground and the program would need to correct it. To correct, it would apply more power to the front rotors searching for a level plane... Would it end up flying a see saw motion through the air? hmmm I wonder. It poses some interesting questions.
I guess if you got it right in applying more power to all 4 rotors, that the lift could be set to conteract the angle of attack towards terra firma.... hmmm
I can't wait for the open board and a bit of additional hardware to test theory on. I see it as a GPS, Gyros, Compass, Accelerometer and Sonar project.
I wonder if there should be a tricky hurdle in the flight, such as a "bridge" to fly under or over. Mock of course.
Title says this is a competition. How is it judged? Time? Accuracy? Combination?
How do we judge successful waypoint arrival? If the quad-copter flies exactly to the GPS coordinates entered and the waypoint is marked exactly (probably not completely correct), it can be 5-10 meters away.
If the quad-copter uses vision to find and navigate to the waypoint marker, must it do all processing onboard or can it send the image to a ground based computer?
John Abshier
Dunno... This entire thread is for fact finding, suggest a fact?
I think time and accuracy mixed would be best.
Cluso99 brought up the idea of landing and using landing distance from target as a scoring point. I kind of like that.
I would vote for all processing done onboard.
The challenge could be complicated by making the waypoint markers smaller, unusual shapes, or removed altogether. To make things more interesting the route could be further complicated by the use of barrage balloons (some lofted to heights not published before competition time), requirements to fly above and below a certain altitude for at least part of the course, etc.
I don't think the arena needs to be 1000' square. In fact, smaller is better. A football field is plenty big for this, because it's precision and not distance that matters.
Obviously manual flight is not the object here, though to meet general FAA regs a human operator must be present, and the competition rules need to be written to allow some minimal human control. In any case, an IMU and associated programming is going to be needed to keep level flight. This is where such a competition gets interesting.
Finally, there should be weight classes, if for no other reason possible regulatory matters.
-- Gordon
At the lowest is GPS-only flight, where one hopes to get to within 25 feet of the waypoints.
At the highest is some form of assisted lighthouse/beacon/marker system (vision, laser, whatever is permitted) where two or more copters must fly in formation from one waypoint to another.
-- Gordon
Not sure I agree with easier about flying robots. More dramatic, certainly!
GPS: When the Xcopter (because they could be tri/quad/hex/octa or any other multi) starts the GPS positions would be given relative to the start positions to account for any satellite variations.
I like the idea of landing at all waypoints. But perhaps one of them could be hover of the ground about 1-2m for 15 secs so we can see how good the hovering is. Obviously this is a subjective part here, but nice to at least see the quality of the hover.
One section between 2 waypoints could be a timed thing.
IMHO the whole event should not be for big prizes, so lots of little prizes. The event should be to see what the xcopters are capable of. Altitude sensing by pressure/temp or OS or some form of ultrasonic sensing or other method. For precision landing, without vision (GPS only), vision, etc.
These days, there is so much that can be done, especially easy with the prop chip. It would be exciting just what can be achieved.
Shame I have been so busy. I have not even had a proper flight yet
I wonder if flight characteristics are outside of what the traditional Robo Magellan competition is about. Besides, with a 9 DOF IMU, and assuming a good airfame like ELEV-8, the hover is going to be rock steady. (Oh, the wind could toss things around a bit, but probably not more than the IMU will be able to handle.) Most of the teams will get 9.5s here, at the least.
The Robo Magellan competition is more about how to autonomously navigate over varied terrain. A bunch of cheap helium-filled balloons can easily serve as the "terrain" to navigate over and through. Pop a balloon and you're down 10 points, or whatever.
Erco, for "easy," ground-based Magellan-type competitions are mostly limited (beyond GPS) to what you can determine in 2D space. Vision is tough when the system is on the ground and looking for recognizable shapes. There are work-arounds, like Northstar type systems when indoors, beacons positioned around the environment, etc., but they all tend to be pretty expensive. A quadcopter has the benefit of being able to determine position from three dimensions -- it's altitude is known (via an altimeter like the one Parallax sells), and armed with an inexpensive camera like the upcoming CMUcam4, the copter could detect obviously painted markers, and use some trig functions to determine location. Although not the same challenge, it's somewhat like being able to get out of a maze by seeing it from above.
So while some things might be simplified, difficulty levels could be put back by adding various conditions to the competition. That could liven things up considerably.
-- Gordon
http://www.youtube.com/watch?v=4cbTF1_jrxs&list=UUtHTG82Nn83T5JSdWdTgkAg&index=1&feature=plcp
Speaking of which, we just watched "Those Magnificent Men in Their Flying Machines" from 1965. The wife & babies enjoyed it as much as I did. Highly recommended!
There's already ArduCopter, a commercially-manufactured quadcopter that does everything a Robo Magellan flying bot would need, except (I think) vision-assist for better accuracy in finding waypoints. Everything is automatic, including take-offs and landings, and an unlimited number of GPS-enabled waypoints. Route software can similate the flight, and it works with Google Maps to show a satellite view of the flight path. It's all open source.
So we're already there, it's just a matter of formalizing some things. I'd also like to see CUMcam4 (which Kwabena should be finishing up this summer) made a part of this project. It's a natural fit.
-- Gordon
Most of those indoor videos of quadcopters doing amazing stuff (including the 8 mini ones syncronized) use an external hi end computer system with a camera array around the room to do most of the computations and also to "pilot" the quads. The quads themselves are essentially dumb RC devices with a gyro (and maybe accellerometer) setup. It's impressive sure, but far less so than if it was actually all on the quads themselves.
Anyone who's done contests (robotic or other) knows that the excitement, sleeplessness and pressure cause people to make dumb mistakes they've never done before. That's why I say they the contests be dramatic. Likely much more fun to spectate than participate in!
Oh, I understand that some are not self-contained, but my point was that stable flight was computerized...somewhere in the chain. The capability is there because of inertial feedback. A contest like Magellan is less about R/C flying and more about programming. In a ground-based Magellan contest there would be ground-based obstacles, and limited view of objects to determine location via any kind of vision system. Completing is actually not at all assured.
In an aerial version, there are no obstacles unless they're placed there (assuming a safe open field of the type the FAA wants you to use), and a teriffic view of the ground and horizon all around. As I mentioned, there's already fully-developed software for autonomous quadcopter flying. You enter GPS coordinates as waypoints, and it calculates how to get there. It requires inertial data integration, but as Chris mentioned above, many quads have that, and those that don't it's not complicated or even horribly expensive to add. Not completing the waypoint circuit would be a bit of a surprise, assuming the copter didn't just fall apart, or a big bird didn't hit it. Or, you neglected to note the barn in the middle of your flight path.
From a pure R/C standpoint, autonomous flights are not as fun for the fliers, because there's no actual flying. You just press a button and watch. Like all things robotics the challenge is in the building and programming.
-- Gordon
I am certianly no expert yet, but what you describe sounds like an EPA GAIN issue. Check page 49 of the HoverFly Users Manual (.pdf) on TUNING GAIN. I could be right... Then again, my nick name is chief dumb thumb, don't blame a crash on me... lol.
In that case I got $5.00 on Kevin Cook!
I thought I would try to find out specifically who was the first to circumnavigate. We know that Magellan did not, he stayed in Cebu. The only vessel of Magellan's fleet that completed the Voyage was the Victoria commanded by Elcano. But who was the first to set foot back on the mainland? He would rightfully be called the very first to circumnavigate. It is unlikely that the captain of the ship would be first so that leaves one of the other 17 men aboard the Victoria.
Pigafetta main have known. Did he record it?
But if the contest is to be named for a great navigator, then I would suggest that it be named after Captain James Cook.