New Products Flash Sale
Jack Crossfire's blog
Posted by Jack Crossfire | May 11, 2008 @ 03:10 AM | 3,336 Views
Tried some 20mph wind. Facing sideways into this wind, 90' rotations of accelerometer derived attitude were common & she often flew away. When not impacted by the wind on the wrong side, she could hover. Also, somehow the camera only shot every 10 seconds & it didn't show the true exposure in viewfinder mode.
Posted by Jack Crossfire | May 10, 2008 @ 03:51 AM | 3,373 Views
Put carbon blades back on & began testing the 2Hz barometer mode. Carbon blades have less coning, less wind influence, less drag, & cost 3x more than plastic. The reduced drag shows up in the yaw trim.

Ran out the transmitter battery in a 1 hour test flight marathon. Basically lived in the air.

We now have a 0.0001 gyro bias weighting. Had 2 anomalies where she entered catastrophic position oscillations in a strong vibration mode of the blades. Reminiscent of a stuck cyclic. Nothing unusual in any flight recording & couldn't reproduce it anymore.

Noticed how noisy she is from the sides despite being very quiet from the front....Continue Reading
Posted by Jack Crossfire | May 09, 2008 @ 03:23 AM | 2,835 Views
With the current wind immunity, logging more automated flight time than any time in the past.

So finally discovered CHDK & installed it. SD cards can no longer be formatted. Downloading from the camera no longer works. The power button doesn't work. Powering up now requires pressing the play button a really long time twice. Part of the screen is covered by a useless zoom & synch widget. The intervalometer scripts don't work. In exchange, U get manual shutter speed.

So today DRDY on the SCP1000 decided to work. According to the multimeter, it puts out 8.5 readings/sec. Sharing the SPI bus with pressure brought the magnetometer down to 17Hz.

The neural network now only calculates horizontal speed. With 600Mhz focused on the horizontal speed, she's pretty resilient in the wind. Pointing downwind is still wild, although she can do it. The tail doesn't have enough authority & it whips around.

Unfortunately, drew a loss of control in a turn in the same place & orientation as 2 previous times. In a certain vibration mode induced by wind & payload, roll seems to drift over time & eventually hit its limit in the moment of truth. Turns near the start of the flight do better than turns after 7 minutes.

Personally don't like Kalman filters. Successful businesses have been built without them. Unsuccessful businesses have been killed by them. They were invented in a time of 9600 Hz guidance computers. GPS assisted kalman filters R still a bit of a science project.

Camera vibration has been a bigger problem since removing the batteries. With this video, we see how effective video stacking can be on a copter. It's like sitting on a table compared to fixed wing video.

More stabilized hovering in daylight (7 min 33 sec)

The autopilot is already getting better video in a hover than we can get in manual control.
Posted by Jack Crossfire | May 07, 2008 @ 11:33 PM | 2,866 Views
Stepped up the collective gain to 0.05 & things seemed to improve. With GPS climb rate, it was 0.02. Pressure climb rate + GPS absolute altitude seems to keep her within 4m in Rain Ramon wind. Even snagged a pirouette without issue.

Once considered impossible, autonomous hover in Rain Ramon wind has become routine. Have no desire to get absolute altitude from pressure, yet. That requires fusing temperature, pressure, & GPS readings.

The real challenge would now be Alameda & Fremont wind. Alameda is to wind what dollars R to inflation, but that gig is now canceled. Fremont is more crowded than Tokyo.

Managed to get some shots of autopilot in really low altitude, with the sky still lit & a background. This position was unflyable before pressure climb rate. Just too many apartments, trees, & hillsides. Now it's as easy as whipping out a credit card & punching up $67.92.
Posted by Jack Crossfire | May 07, 2008 @ 03:15 AM | 3,221 Views
So voltmeter #1 got fried when it was used on a fluorescent ballast. Voltmeter #2 sucked & finally had a probe fall off. Voltmeter #3 is the best, so far. RadioShack has been raising prices & making everything suck more. Not so at Sears, where they finally got China to make a decent voltmeter.

The display is unreadable except directly overhead. The cont noise is virtually silent. Most functions R buried under several select button presses. At least it has capacitance, oscillator, & duty cycle modes. It senses VicaCopter's 40Mhz heart despite being rated for 10Mhz.

Turns out the DRDY on the SCP1000 isn't needed. It constantly spits out the same reading over & over between updates.

The answer is no. Rotor wash doesn't affect the pressure reading. U can blow on it as hard as U can & it won't deviate.

Barometer climb rate works. Seems a bit more stable in the wind & can probably stand higher gains. The gains R still identical to GPS. Absolute altitude is still GPS. Only climb rate is pressure.
Posted by Jack Crossfire | May 06, 2008 @ 03:35 AM | 3,220 Views
Since a barometer was on the manifest for VicaGlider & the credit card reset on May 1, decided to nuke it.

So far, the DRDY pins R not tristate. U need discrete microcontroller inputs for those in addition to the discrete outputs for the chip selects. Those chip selects need to be diode/resistor converters so they default to high.

We have the barometer & mag sharing the same SPI bus, but the barometer updates far less frequently than the mag. U have to send some commands to the barometer, wait 0.5 seconds for it to send an interrupt, get byte #1, send more commands, wait, get byte #2, & so on. Getting around that problem is going to put this story out of commission for a while.

Oh and yes. Figured out why Maker Faireeeee looked so washed out. Been raising the black levels because everything looks too dark on the HP.
Posted by Jack Crossfire | May 04, 2008 @ 01:24 AM | 3,225 Views
So the 15mm fisheye was natural on the 20D, but on the 5D it's unbearable. Finally wrote the fisheye correction for Cinelerra. Our preferred mode is to straighten just the vertical lines & let the horizontal lines bend although the filter can do a strict conversion.

The filter is much faster in OpenGL, but OpenGL maxes out at 4096x2048 & uses an imprecise lookup table for atan. Making the crooked straight is just a matter of atan.

Did some more hover timelapses in strong wind. Static hovers seemed possible. Position hold was once again lost when turning.

Evening autopilot in high wind. (0 min 51 sec)
...Continue Reading
Posted by Jack Crossfire | May 02, 2008 @ 12:13 PM | 3,319 Views
Regarding VicaGlider, now thinking if we moved on it, it would get a 200Mhz Gumstix with a VicaCopter load modified for fixed wing. Would also invest in the full IMU, pitot tube & barometer that everyone else uses. There will be no flying wing.

Got our first successful night flight with video & turns. No wind of any kind. The camera was aimed too low, cutting off the horizon. Got a decent view of the shopping center from altitude. Applied video stabilization in software. Static hovers are much choppier than the constantly moving videos most people make, but static hovers are harder.

Night flight 5/1/08 (5 min 13 sec)

Flight time with the 4.1Ah battery, camera, & camera battery eliminator was 12 min. Total flight time on the 600Mhz is 4 hours. Autopilot time is 3 hours....Continue Reading
Posted by Jack Crossfire | May 01, 2008 @ 01:06 AM | 2,909 Views
So been flying in higher wind to study the performance characteristics. The current software chops through local wind much better than any previous software and yes, it's using all 600Mhz.

The problem is altitude is lost suddenly if the wind stops or altitude increases forever if the wind increases. Constant wind of very high speed is probably accessible if she points into it.

Also, attitude hold can be lost if the tail experiences too much wagging. Point downwind & things become very choppy. Point upwind & she's quite stable.

Commanded 3 pirouettes in wind. All of them had sudden altitude losses. The first drop required an abort. On the next 2 we stopped if she started losing altitude. Also have evidence of GPS drifting 4m.

The waypoint flight mode stops turning if she deviates outside the satisfy radius. The operator directed mode requires user judgement.

10 minute flight times R a real drag. The next steps would be $600 for carrier phase GPS, a gas airframe, $80 for 900Mhz video. The economic wall for the next step is finally impassible but won't be someday. The latest source code is on
Posted by Jack Crossfire | Apr 29, 2008 @ 03:17 AM | 5,181 Views
Discovered the idler pulley can be removed. Without it, we supposedly have less vibration & energy loss. Mainly, it's one less gift from Align & the less Align gifts the better. $10 of landing gear & 1 less idler pulley later, she's flying again.

Operator directed flight (0 min 20 sec)

That was 9 min of playing around with operator directed autopilot, forward flight, & turns. Results in wind will be much worse, of course.
Posted by Jack Crossfire | Apr 27, 2008 @ 11:00 PM | 7,390 Views
So had a descent anomaly followed by a hard roll at 5ft, all under autopilot. Nothing survived that crash, not even bacteria. Observed a hard roll to port. The IMU sensed a hard roll to starboard.

G forces dropped during the descent anomaly. When G force crossed from + to -, the accelerometer result flipped 180' in a 0.04 second period, causing the blended result to show a sudden roll starboard.

How do you sanity check the accelerometer? You can't check total magnitude.

Now the stabilized footage from our last daylight mission.

Stabilized hovers (0 min 45 sec)

Posted by Jack Crossfire | Apr 27, 2008 @ 02:45 AM | 5,095 Views
Managed to get off the ground & land safely with the camera & Russian Heroine on the scene. She did autonomous hovering briefly. Unfortunately high wind seems to have killed off opportunities beyond that. More notably, the IMU seemed to get lost & couldn't return to the neutral attitude after a certain point.

All the autonomous hovers were too high to see the pilot. The only photos with pilot & Russian Heroine were manually obtained.

Finally, one of the XCite batteries was never charged, so we got only 1 flight.

One observation is that she can probably handle straight line wind but the kind of wind in this location reverses direction constantly....Continue Reading
Posted by Jack Crossfire | Apr 25, 2008 @ 05:34 AM | 7,308 Views
Well, U can look at falling prices for carrier phase GPS and say all this neural network hacking could be eliminated with money & replaced with a bulletproof system.

That $300 board needs another $300 to do 10Hz. Last year, the Novatel was doing 20Hz for $1000. Electronics drop 1/2 their price every 2 years. We're probably 2 years away from anything coming close to replacing the neural network & only if we still care about it. 4 years away from it being economical for after-tax pilots. The neural network is going to be around for a long time.

Now IMU's R rapidly becoming ubiquitous. They'll start using 324 pixel mouse cameras to add position hold to indoor dual rotors because that's the easiest money.

What happened to VicaGlider, you wonder. Well, it became less & less a reuse of spare parts & more & more a second parallel money pit of replacement parts. The lack of open space was another killer. The T-Rex hasn't had the endless downtime waiting for parts that the Corona had. Finally, there's the day job.

The 622Mhz computer now has 177 min of flying & 133 min of autopilot, a ratio we'd expect. It hasn't reset in flight like the last one did. Suspect replacement of the power connector fixed that.

The FIR neural network started working as expected. Altitude seems a bit more stable.
Posted by Jack Crossfire | Apr 24, 2008 @ 06:54 AM | 5,100 Views
Somewhere in the testing since 11/07 the navigation went from a FIR neural network to current inputs only. The FIR method is quite hard to verify, reduces training to 4275 evolutions/sec but used to give slightly better results than current inputs only.

Gave up on plotting graphs of velocity prediction runs. Attitude hold has gotten good enough that it always defeats the predicted velocity before it happens. The only way to compare these things is to fly.

Sadly, haven't gotten the FIR implementation to work in our Rain Ramon visits. Got some more words painted using the tried & true algorithm. Vimeo has 8 hours to go on processing that one.

Well, 4/22/08 came & went & no EOS 5D Mark II appeared. Haven't used the 100mm lens much so here R some shots.
Posted by Jack Crossfire | Apr 21, 2008 @ 01:45 AM | 5,180 Views
Had exactly 1 day off since Heroine Cam arrived. Once a company begins 6 day work weeks, getting back to 5 day work weeks is a matter of either getting into a low profile project or quitting. Once acquired, the taste for winning more contracts by leveraging employee personal time rarely gets sour.

Path through 35mm (0 min 40 sec)

Flight projection through the 35mm sensor followed by projection in Google Earth....Continue Reading
Posted by Jack Crossfire | Apr 19, 2008 @ 03:04 AM | 5,115 Views
Decided to finally try a washout filter instead of a neural network. Recommended by none other than Major League Baseball himself.

The washout filter was much looser than the neural network in stable air. Maybe it could be tuned, but we're after major improvement & clearly that isn't washing out.

Got better focusing by focusing on the tail boom & setting the flash to stop down -1. It's beginning to look like an extremely narrow depth of field, very small pixels, combined with a lack of the focusing assist range we had on the 20D. The IR assist just doesn't cut it & there's no way back to flash assist. It's clearly focusing in front of the object.

Would now agree the 5D sensor is slightly less noisy at ISO 1600 than the 20D sensor.

Finally hovered out a battery. The flight time with a brand new 4.1Ah battery in ground effect is 16 min.

The day job is requiring work on weekends. It's one of those jobs where it's more tradition than necessity. We have probably only 10 hours in Rain Ramon per week now.
Posted by Jack Crossfire | Apr 18, 2008 @ 03:54 AM | 5,081 Views
Clearly the swashplate neural network has nicely decoupled the 3 swashplate axes since the long coveted single module for navigation made no difference. We're looking for significant improvements here, but all the designs seem to have milked the last bit of information out of 1 Hz GPS.

There's a slight chance adding MEMS acceleration to the 3 module network improved its altitude hold. More evolutions to grind than the 1 module network.

Would be nice if the software could automatically throttle the evolutions to hit an exact CPU load. It's the times it gets it wrong that we're worried about.

1 module, 30 sec history
current GPS acceleration
commanded tilt
predicted GPS acceleration

same as 3 modules

3 module, 30 sec history
current GPS acceleration
current MEMS acceleration
commanded tilt
predicted GPS acceleration

Same as without MEMS.
Significant oscillation during training.
Possibly better altitude hold.

3 module, 60 sec history
current GPS acceleration
commanded tilt
predicted GPS acceleration

Same as 30 sec.
Longer training time.

So much for that....Continue Reading
Posted by Jack Crossfire | Apr 17, 2008 @ 01:05 PM | 5,294 Views
Got lucky with the crash damage. Only $25 & 5 hours of repairs. Remarkably, the main shaft didn't bend. Got the computer running again. The tiny hirose connector was predicted to break in it's first opportunity, & it did. Now we know where to pad the computer.

Got the serial port reading SiRF binary, 9600 baud, & low latency, thanks to Canadia. U need to set c_cc[VMIN] to 1 & c_cc[VTIME] to 1. Now we know the binary protocol gives much more accurate readings than NMEA.

Discovered if U have multiple BEC's, only connect 1 of the ground lines. Reversing the polarity of the computer tap causes a short circuit between +12V from the ground & +6V from the BEC. That's what melted a wire before. Now the remaining unmelted RCE-BL35X wires are 6V & PWM. The computer's diode protects the +12V line but not the ground.

The answer is yes, using SiRF binary protocol gives much more accurate readings than NMEA. Got good results using the 3 module network. The 1 module network was disappointing. Still have more combinations to try before declaring the 622Mhz a waste of money.

Web 2.0 is getting back to full power after a few months of mass layoffs, & lots of middle managers R getting promoted. The bigger the manager, the smaller the car. Spending almost all day in traffic. Monster accidents shut down the freeway all morning. More accidents stopped traffic all night. Only getting...Continue Reading