HobbyKing.com New Products Flash Sale
Jack Crossfire's blog
Posted by Jack Crossfire | Mar 19, 2008 @ 06:18 AM | 4,662 Views
Well, after banging on it, decided the easiest solution for the cyclic crosstalk was to make longer pushrods but a metric tap costs big bucks. We're trying to use lower head speeds than normal to increase flight time.

Other than that, the other open issue is the new roll oscillations. They seemed 2 B a new artifact of the neural network, so reduced the neural network gain & the oscillations didn't go away. Back to attitude hold tests.

Altitude is still solid with the cyclic tuning. Have not had a single vortex ring state in 2 days. Amazing how invisible cyclic crosstalk is to a human yet the computer can't handle it.

Did the long mission with only 1 90' turn. 4 minutes with the turn. With the downgraded neural network gain & 4m/s velocity targets, she blew threw that mission like Hillary. The safety pilot got destroyed trying to keep up with her. The neural network didn't over correct as much & the route was sloppier as expected.
Posted by Jack Crossfire | Mar 18, 2008 @ 04:58 AM | 6,134 Views
The answer is yes. There was crosstalk between cyclic & collective. 1mm of down collective in maximum roll 2 B exact. While we can't eliminate the crosstalk, turned it into 1mm of up collective.

Despite the doubters, the proof was in the test flight. Maybe it was the weather, but no vortex ring states & the tightest altitude of any mission. CCPM crosstalk definitely seems to impact results.

Unfortunately, the sprinklers R coming on again, so we're watching copter & sprinklers simultaneously.

Eliminating all CCPM crosstalk mechanically seems 2 B impossible. The only way would be to measure the exact dimensions & mathematically model the pushrods, horns, & swashplate. Then there's replacing the swashplate with a solenoid. Then of course, U can have the computer set every possible servo position & photograph the swashplate, storing every input & position....Continue Reading
Posted by Jack Crossfire | Mar 17, 2008 @ 04:01 AM | 4,895 Views
Reduced the turn rate to 4.5 deg/sec & that basically ended drifting during turns. It's not as fast as the industry standard but this is an after-tax operation.

With such slow turns, the mission came to 5min. Previously it was around 4min30sec. The advantage is more chance of actually finishing. Pretty confident, given enough altitude & dead calm air, she could perform a mission beyond visual range....Continue Reading
Posted by Jack Crossfire | Mar 16, 2008 @ 03:51 AM | 4,733 Views
So the GPS satellite launched today does L1, L2 signals but no L5. There's now a choice between sonar, laser range finding, computer vision, & just waiting for better GPS accuracy.

1575.42 MHz
Satellites: Block I 1978, Block II
Completion: 1995

1227.60 MHz
Satellites: Block IIR-M 2003
Completion: 2013
Value: dual-frequency ionospheric delay correction with L1. Makes DGPS obsolete

1176.45 MHz
Satellites: IIR-20(M) June 2008, Block IIF, Block III 2013
Completion: When IIR-M satellites reach end of life and are replaced
Value: 6 dB higher power, more bandwidth, longer code, 20 dB higher power
spot beams for Calif*

3cm, fast velocity readings from carrier phase will never reach consumer level because it's not used anywhere.

Next, we have software revo mixing & software trimming. In the name of the tightest heading possible, added software revo mixing to the computer. Pumping collective is definitely a bit easier with revo mixing. It doesn't lock heading, but it greatly stabilizes the yaw. 3D masters have briefly dropped that they don't use heading hold gyros, but not often since they're paid to sell heading hold gyros.

Global equilibrium doesn't seem to work. In a very long 11min flight, she built up a positive roll to counteract strong wind. Beginning the counterclockwise turn, the positive roll should have turned into positive pitch but instead stayed a roll & moved her way way out west.

Now leaning towards a play between the IMU tilt & the equilibrium tilt.

> North-south surface gradients have been increasing after upper trough
> passage and gale force gusts are currently being reported at the
> offshore buoys.
> Winds have begun to pick up in the hills and should
> continue to strengthen at the higher elevations overnight

That's why this problem is taking a long time to track down.
Posted by Jack Crossfire | Mar 14, 2008 @ 06:43 AM | 4,840 Views
Well, decided the Maxbotix isn't robust enough, there aren't enough weather windows to play around, so it won't even be giving a sonar hover demo.

The idea of a custom sonar velocity sensor is still intriguing. The Maxbotic transducer goes up to 117dB using 20V. The strategy would be to buy another Maxbotic transducer & transmit a wave on it full time using a custom circuit. Then we would use the existing board in full time receive to detect the returning wave's frequency. The filters would be powered up but the output would go to an alternative PIC.

So much for that. Time for timelapse documentation of some autopilot missions. 1 used ISO 400, 2sec, F2.8. 2 used ISO 1600, 1/250 sec, F2.8. 1 was a sideways dolly, 180' turn, sideways dolly, & victory hover. 2 was a static hover. Note how the position changes as GPS drifts over time.

The batteries could not do more than a handful of flash photos before dying, & the shutter computer didn't work, so we were swapping batteries every few seconds, manually opening shutter, & standing by on the autopilot abort.

Flight 1 was 10min. Flight 2 was 11min & battery exhaustion.

Autopilot timelapse (0 min 45 sec)

Posted by Jack Crossfire | Mar 13, 2008 @ 04:41 AM | 4,591 Views
The answer is yes, if aggressively filtered in software, isolated from rotor wash, the Maxbotics EZ1 could probably hover 1-2m off the ground.

If it's on the ground, too high, or it doesn't receive a reflection, the Maxbotics outputs a maximum altitude. Most of the time it just outputs a random glitch & it's up 2 U to detect it. On 5V, it simply doesn't generate enough sound pressure to cut through engine noise & reflect off grass.

No-one wants to hover 1-2m off the ground because of decapitation risk, though it would make a nifty demo. It could enforce an altitude floor. It would require a heavy enclosure on the tail boom to isolate it from rotor wash & achieve the necessary calibration height. Mike Fouche's $17,000 beasts don't even use sonar.

The Maxbotics EZ4 might reject engine noise better, but we've played this game before. There's no such thing as a free lunch. It looks like it rejects noise but has shorter range.
Posted by Jack Crossfire | Mar 12, 2008 @ 03:46 AM | 4,977 Views
Well, got the Maxbotix EZ1 integrated & the fun began. First impressions R that it can't handle the vibrations or noise of flight. Got slight glitches in a handheld test too, but clearly flight is out of reach for the Maxbotix.

Sad, considering the software for sonar guidance was already written.
Posted by Jack Crossfire | Mar 11, 2008 @ 04:47 AM | 4,256 Views
Found the ideal target V 2 B 4m/s in the forward flight mode. That got real velocity between 0 & 2m/s, the fastest Americans can walk before they fall over.

After 5 tries, finally got her to finish the long distance mission. The mission took 8m50s & total flight time was 11min. Altitude hold with this mission was pure luck.

Set waypoint 3 2m lower than the others to keep her in visual range. Altitude hold is so bad with waypoint #3, suspect a weather change in that area. Had a huge drift with the turn. Also had a bug where target heading suddenly snapped to waypoint heading.

Got gas today even though we didn't need it. There definitely were more cars than usual but not the 70's lines. People R starting to figure out they can save 20% by keeping it over half full, despite the internet hypnosis machine.
Posted by Jack Crossfire | Mar 10, 2008 @ 04:45 AM | 4,461 Views
So now we hold target speed at a certain large number for forward flight & scale it for hovering. The mode is determined by the waypoint file & there's a complicated rule for turns.

If SPEED=0, the hover algorithm is used to track the waypoint. If either current waypoint SPEED or next waypoint speed SPEED is 0, the hover algorithm is used to make the turn. If SPEED is >0 for the next waypoint, the target speed after turn completion is held at SPEED.

With missions this long & only 2 batteries, only get 2 flights per commute.

Unfortunately, on flight 1 with SPEED=2 the battery died after finishing half the mission. Accidently had the final 1/3 of the mission in SPEED=0 & she crawled to the final waypoint. Flight time: 12m.

On flight 2 with SPEED=8, it was faster than the safety pilot could run. When she hit waypoint 2 she dove for the ground, entered altitude oscillation & couldn't recover. Looks like she barely missed it & tried to change direction suddenly. Without adapting to the new direction, she ended up stationary & couldn't handle the loss of translational lift.
Posted by Jack Crossfire | Mar 09, 2008 @ 05:41 PM | 4,114 Views
Before the next monster flu set in, got some autonomous hovering video in high wind. Given enough space, she can actually do it without sonar altimeter.

As Silicon Valley real estate inflation gets higher & higher & higher, the offices get smaller & smaller & the sick time gets worse & worse. Soon every company will share the same office & use it for 1 hour a day. Cisco will use it at 1am. Adobe will use it at 2am. Goog will use it at 3am.

Autonomous hover in high wind (1 min 34 sec)

Posted by Jack Crossfire | Mar 08, 2008 @ 06:03 AM | 2,666 Views
The previous algorithm created target X & Y velocities by scaling X & Y position errors separately. Now we convert position error to a vector & make target X & Y velocities a scale of the vector.

The result: straighter paths, fewer PID constants, tighter hovers & more tree clearance. Velocities R slower because she's sharing all 2m/s between X & Y.

Took 7m48s to complete the mission. That was followed by 3m9s of autonomous victory hover until battery exhaustion. 11m16s of total flight time on the 7 month old 3.3Ah 11.1V. Mind U, this was a longer mission than the last one & speed will always B limited by safety pilot sprint speed.

Spark No Fun is the first store we've seen which lines up your entire order history in gory detail, representing most of VicaCopter's autopilot cost.

$28.25 Sonar
$27.34 Active GPS antenna
$24.34 Passive GPS antenna
$224.77 ADXRS150 gyros
$112.42 EB-85 GPS
$38.27 ADXL330 accelerometer
$158.67 Micromag3, EM-406 GPS, LIS3LV02DQ accelerometer

Rough calculations for the rest.
PIC spares $40
Gumstix $140
uBlox GPS $140

Since the alternative to XBee & radio hacking was a $350 radio set, that's omitted. The autopilot alone has been under $940. If we didn't follow so many dead ends, it would have only been $600. It's the common RC stuff which has been slaughtering us.

LiPo batteries, servos, airframes: $1,000,000,000,000,000,000,000,000,000,000,000,000 ,000,000,000,000,000,000,000
Posted by Jack Crossfire | Mar 07, 2008 @ 04:38 AM | 2,901 Views
Had VicaCopter's best performance so far on the test mission. 4m45s Turns were tighter though still not European quality. The neural network previously used commanded attitude for training. Moved the neural network to use detected attitude for training & commanded attitude for prediction. That may have degraded performance, but also had a slight breeze.

Also fixed a bug where IMU alignment was double subtracted from neural network training.

To cover longer distances, really need a waypoint mode which doesn't reduce speed as the distance closes. Currently speed is reduced as per the hovering algorithm. Entering missions is really tedious. Really need a way to plot markers in Google Earth & convert that to a waypoint file.

Finally got nighttime closeup video of autonomous hover. The A560 is no TX1 for low light.

Nighttime autonomous hover closeup (0 min 52 sec)

Posted by Jack Crossfire | Mar 06, 2008 @ 04:36 AM | 3,135 Views
The Airtronics 94761 is definitely the most effective tail ride we've ever seen. Software heading hold was seriously unreliable & it probably won't B necessary with the Airtronics. It almost locks heading in rate mode alone.

Also in this issue, used the neural network to predict position for the first time. Predicted position + Airtronics + dead calm air gave the fastest + tightest mission she's ever done. 4m41s Did not observe any obvious vortex ring states.

Unfortunately, once again seemed to encounter saturated cyclic integrals in the victory hover & making turns still requires too much room for Calif*'s limited space. Turn drift & victory hover oscillation may all be from integral saturation.

The sonar altimeter is on the way anyway. The trick with the sonar is it maxes out 25ft above ground & there is no flat ground in Calif*. It needs a manual mode selection.

Mode 1: Full time GPS altitude. Use absolute height above mean sea level. Switch to sonar when sonar detects less than minimum sonar altitude & lock to minimum.

Mode 2: Full time sonar altitude. Lock constant height relative to ground.

Obviously path.conf needs another column for sonar altitude & settings.conf needs a sonar mode.
Posted by Jack Crossfire | Mar 05, 2008 @ 05:19 PM | 4,784 Views
When U lightly bump a Taiwan-Rex it's $3 for a new main shaft. When U drop a T-Rex on concrete, it means 2 commutes of repairs, many trips to the hobby shop and $60. That yesterday's crash was on a narrow strip of concrete in 1 acre of grass was pure luck.

So far:
New main gear $2
New rotor head $33
New blades $16

Since we used plastic on the Corona 4 a year, & plastic hasn't been a problem for the T-Rex, there will be no more carbon blades.

Yaw anomalies have caused 2 of the last 3 crashes & suspect the Futaba S3102 may be too slow for Taiwan-Rex. To attack the yaw problems, now have an Airtronics 94761Z on the tail, the most powerful micro servo available locally.

Now have enough spare servos to build an EasyGlider using all micro servos.
Posted by Jack Crossfire | Mar 04, 2008 @ 05:45 PM | 2,591 Views
The autonomous, flybarless, T-Rex 450 hovers for a moment before crashing.

The actual damage was closer to $50. As more & more functions are moved to software, the crashes get more spectacular. This time had a sudden loss of altitude, routine collective up to escape from autopilot, sudden pirouette at low altitude, crash.

Software heading hold is too unstable for fast collective changes. Software rate damping is too loose. Using a lower head speed than normal & relying on software to increase flight time, but with a very slow Futaba S3102. The next step is a more expensive servo.

Any wind just a hair above dead calm causes loss of altitude control,
but position seems more stable.

Autonomous hover followed by crash (0 min 53 sec)

Posted by Jack Crossfire | Mar 04, 2008 @ 05:29 AM | 2,911 Views
Found a bug in the neural network where collective was never calibrated for the T-Rex. It fed 0 for all collective. With that fixed, hovered out the kit battery with much more stable altitude. Mind U, the kit battery now only flies for 7min & GPS drifted a lot.

Well, a sonar range finder for VicaCopter is probably going to happen. U can write dozens of pages about how flight should work, but it's no substitute for functional sensors. Though only intended for more stability, it'll probably end up turning into auto takeoff & landing.

While Heroine is back online with a new power supply, we now spend 75% of our time in Rain Ramon just keeping equipment alive.
Posted by Jack Crossfire | Mar 03, 2008 @ 05:28 AM | 3,652 Views
Had our first software crash in flight. Reached the final waypoint and began oscillating as before. Disengaged autopilot, put her on the ground, and commanded engine cutoff. The engine did not stop & lost the computer. Thrashed for 3 seconds until the PIC detected a fault & killed it.

Could not reproduce the crash on any previous simulation, but moved back to plastic blades anyways. May have been static electricity when touching the ground. The neural network now stores & recalls the weights on shutdown & startup. The idea is to simulate the exact same feedback, which is not possible with random weights.

On the final attempt before the next commute, she did finish the mission after 8m30s but entered another oscillation after finishing.
Posted by Jack Crossfire | Mar 02, 2008 @ 03:37 AM | 2,901 Views
Have found KML extremely limited. U can't construct very complicated objects out of line segments & Place Markers. There's no polygon sorting. It's more intended for annotation than virtual reality despite many a Google fanboy.

The things we miss most R anaglyph support & getting more precise coordinates for markers. Of course, it crashes too much 2 B very useful.

The latest flight program requires position to be in the satisfy radius to perform a turn. If position deviates outside, the turn stops.

Also increased collective P gain back to Corona levels.

Finally got the 180' turn mission to work, after turn rate was reduced & wind briefly calmed down, & the software was fixed. Google doesn't have very accurate elevations. The flights in our plots R fudged to appear above ground.
Posted by Jack Crossfire | Mar 01, 2008 @ 07:36 AM | 2,675 Views
west pi wrote:
> strong winds likely to persist in the north and East Bay
> hills through Monday

3 years & Heroine's power supply finally died. $100 doesn't buy what $25 used to. Fortunately, we still have 2 computers & that ain't bad.

Also, the 30" HP doesn't work on the Quadro FX 4000 until X11 boots. If anything is corrupted on the hard drive, U can't get a display at all.

Altitude hold with the software climb rate was definitely more stable than doppler shift, even in lousier weather. Hovered a battery out with some heading changes in formidable wind. 11m14s under computer control. She touched the ground a few times.

Seeing a slight improvement with world orientation tracking. 90' turns R still the limit but she seems a little more stable. 180' is still a matter of luck. In very unflyable weather, managed to get a 180' turn out of her before wind sent her to the ground.

The problem is IMU alignment contributes a constant offset to copter orientation & it must be added & subtracted from the navigation in bug prone ways.
Posted by Jack Crossfire | Feb 29, 2008 @ 06:09 PM | 4,961 Views
This is the last simulation of orientation feedback in copter frame, because now orientation feedback is in world frame. Have found the feedback in different simulation runs is different every time. The neural network is seeded with random weights before every run & it may actually remember starting information through entire flights.