Posted by Chris Anderson on September 20, 2008 at 11:30am
Dean Goedde's highly-anticipated Attopilot, in Pro, Basic and RTL form (estimated costs: $800, $400, $100), will have a proper home in ten days, when his site goes live. [UPDATE: Dean doesn't have an exact date for availability, but as he notes in the comment below, he expects to start taking order in 4-6 weeks, with units shipping around that time.] Lots of background info in this incredibly long thread.
Ah, i thought mismapping. For me, the IMU takes care of that. Then it expects gives an expected Lat/long for the next sync. Ofcourse the noise and encryption makes it only good to a few yards X and Y. It is a redundancy, the only redundancy. If i lose everything, and have to reboot in mid air, i'm not exactly sure how to handle that one. I have an idea though lol.
I guess my biggest question then, in terms of contingencies is how you handled it missing a waypoint and how did you keep it form turning around and around, esp if the craft is high speed, ie, a jet one. I mean, if you have a craft going at 60mph with a 1hz gps (unless you're using a 5hz), you can miss the waypoint by a few meters on the next sync. does the craft assume it hit it, or what? At 100mph, it would probably be a good bit worse. Then again, i don't know anyone using a jet engine based UAV lol, but i only know like a few guys who have anything more than a glorified FPV, and all are on this forum. lol.
By contengencies I mean what to do if GPS lock is lost, or GPS dies, or a sensor fails, or gives data that is outside of a reasonable range... I think your Kalman filter can do these things as far as reasonableness of the data goes... but the GPS lock lost is important to handle, and is not part of a data filter, per se.
hehe you def have my admiration. the handling contigencies i am actually trying to code at the moment. With an IMU and a 1hz GPS, i can only do so much. I'm thinking of using an appropriate kalman filter with a response threshold in order to making it not over compensate. I want to say that one day i may join you in building an autopilot, but the way mine is going right now...hehe, i ask myself sometimes why you probably did at first- what was i thinking, believing that i could do this damn thing from scratch? You have given me hope though.
My biggest hope is to get a processor on there that is powerful enough to require no base station intervention. Once i get that...hehe...yeah...good times after that.
"..This is an easy project in theory.." - I might have thought that at the beginning, but listing out what it does shows that even at first impression it is truly like rocket science! Atto is juggling stabilization, mixing is possible for elevons and V-tail, navigation, data logging, voltage and current and power and mAh consumed of the propulsion pack, handling of contingencies, triggers for payloads, 2-way telemtry including proprietary nmea data stream to the ground for moving map software and virtual cockpit, scaling of servo response based on airspeed, scaling of elevator response based also on roll angle, some 1 dozen real-time data filters that truly calculate least-squares fit of large data sets, and all that is just the coding! Then there is the hardware and PCB layout aspect of it all... choice of the best and right components, and you can quickly see why this sort of thing is the topic of PhD theses, and very few people are doing this - the making of new autopilots. Atto is so complex, it has taken a 1-year Beta testing effor of some 17 testers, and me to quit my day job to do this full time. AttoPilot actually has more computation power than the MicroPilot MP2028, 4x more than the U-Nav3500, and some 40X of the PicoNAT. I am using every scrap of computation power I can harness. Atto V1.7 is a 10x improvement of the performance/price ratio of systems out there ready to go (as opposed to DIY like Paparazzi), and I think Atto will be a disruptive advancement to other companies selling units in the sub $10k market, especially when I release the 9DOF IMU version. The current thermopile versions are intended to bring HUGE performance to the sub $1k market... the market previously owned by the only game in town, the Pico series.
Right - cheap one is GPS only, though it is specifically setup to work with FMA's CoPilot system which sells for $70 or so. The $100 Atto RTL is NOT a programmable autopilot, though. The Arduino autopilot might be what you want for that.
The $500 and $800 Atto units can best be thought of as an advanced 3-axis autopilot with economy of thermopile system in 3 axes, PLUS an Eagle Tree data logging system all rolled into one, with the added benefit that Atto actually uses all the data, rather than just logging it and beaming it to the ground.
chris - i like your idea and you are very smart to come up with a cell phone based project i sawthat andi was stunned, if you could please send me the plans and details about all the electronics and details invilved in building that uav aircraft would be highly greatful,
cheers!:)
Comments
I guess my biggest question then, in terms of contingencies is how you handled it missing a waypoint and how did you keep it form turning around and around, esp if the craft is high speed, ie, a jet one. I mean, if you have a craft going at 60mph with a 1hz gps (unless you're using a 5hz), you can miss the waypoint by a few meters on the next sync. does the craft assume it hit it, or what? At 100mph, it would probably be a good bit worse. Then again, i don't know anyone using a jet engine based UAV lol, but i only know like a few guys who have anything more than a glorified FPV, and all are on this forum. lol.
My biggest hope is to get a processor on there that is powerful enough to require no base station intervention. Once i get that...hehe...yeah...good times after that.
i wanted to congratulate you as well. This is an easy project in theory, much harder in practice. I hope you have great success with it :).
The $500 and $800 Atto units can best be thought of as an advanced 3-axis autopilot with economy of thermopile system in 3 axes, PLUS an Eagle Tree data logging system all rolled into one, with the added benefit that Atto actually uses all the data, rather than just logging it and beaming it to the ground.
cheers!:)
-
1
-
2
-
3
of 3 Next