New Products Flash Sale
Jack Crossfire's blog
Posted by Jack Crossfire | Aug 06, 2014 @ 10:17 PM | 2,347 Views
So there was an interview at a follow cam startup. At 1st sight, the guy made up his mind. There was a courtesy lunch & then what was originally planned as a 5 hour tour was terminated after 1 hour.

They weren't interested in anyone over 30, but there was also a large experience gap. They were all just starting out in UAV design & may have wanted the novel ideas that people with minimal experience give, rather than the clouding of years of bad experiences. They may have wanted absolutely no skepticism that the idea was going to work.

They were way behind the other follow cams. Since everyone relies on the same 3D Robotics load with minor changes, it probably wasn't a big deal to languish. Like most every other idea getting the most funding, it was very unlikely to work. Would still say there is enough time before the sh*t hits the fan for the follow cam startups to make significant progress towards the goal.

It was disappointing that a place with no prototypes, minimal experience, & a very hard idea to realize was able to get significant funding, while the other ventures in toys & video editing were unable to get any funding after many prototypes, massive experience, & despite ideas that were equally feasible.

The interview guy had a PhD in CS from the U of T at Austin. The CEO had an MBA from Stanford. The education is definitely a consistent factor among those who got funded, while the unfunded consistently had very little formal education.
Posted by Jack Crossfire | Aug 06, 2014 @ 12:36 AM | 2,654 Views

After running out of IP addresses, Comca$t disabled all the old routers. After a few days without service, an upgraded router with private subnet appeared. The upgraded router was almost able to saturate the home network, but completely unaffordable. It was surprisingly the same speed as the $100 plan, but they charged only $88.

It was finally time to start experimenting with shutting down the cable guy.

3 megabit plan, for $73 instead of $88. It's manely a test to see if it can be done away with.

...Continue Reading
Posted by Jack Crossfire | Aug 04, 2014 @ 01:12 PM | 2,000 Views

With the growing amount of red on the war map, it's amazing the economy hasn't been affected. The oil keeps flowing & the stock market keeps going up. As long as you don't fly over 1/3 of the world, travel is unimpeded. If conflicts in the last 5 years are all counted, the entire map is red.

It would be convenient if the news provided a summary of all the fighting or referred to it all as "the war", but it's not officially a world war. It can't be officially called "the war" because US's president is completely in over his head.
Posted by Jack Crossfire | Aug 03, 2014 @ 06:55 PM | 2,316 Views
Despite the media focus on, massive spending on CS in primary school, massive expansion for H-1B visas , there's no demand for people who can just program. There is demand for people with formal CS education & massive experience in very specific languages like Hack, Go, & Swift, languages invented by 1 company specifically for 1 application.

It's a new way of doing business. Languages based on industry standards, that could be transferred between many jobs are all but obsolete. Each company now develops their own language, specifically for their needs. Google invented Go. Apple invented Swift. Facebook invented Hack. There may soon be a time when every single application begins with developing a new language.

The key to the 1 language per application model was eliminating the need for every new language to have new libraries for accessing the system. Instead of requiring new libraries, all new languages access the system through HTML. They just communicate to the system through a character stream & might have some libraries for parsing the DOM tree.

A positive sign is while interpreted languages were the rage from 2000-2010, all the new languages are natively assembled again. It's more secure to have the output of a trusted compiler on the development machine sent directly to the hardware on the consumer machine, than rely on an interpreter on the consumer machine which could have been compromised. Developers have famously breached the Dalvik...Continue Reading
Posted by Jack Crossfire | Aug 02, 2014 @ 07:11 PM | 2,287 Views

No, it doesn't work, but it has been willed into working by massive numbers of reposts saying it works. It's a fine example of how very little of the internet is real, from the economic boom to the self driving cars, but has been willed into reality by the number of reposts. People are happy about the illusions, which is probably good enough. Someday we may never need know a day outside a simulated world.

Read about the quantum vacuum plasma thruster years ago & it wasn't any more credible then with an audience of 1 than it is now with an audience of 50 billion. The theory is by bouncing microwaves in a container, some of them move forwards & some move backwards. The mass of the photons in the microwaves creates inertia. Since 1 end reflects fewer microwaves than the other, more photons push in 1 direction & since the waves reflect many times, a single wavelength produces many bounces.

Of course, it does rely on expelling mass like every other engine. The mass of the photons leaked from the end with lower reflection is what propels it forwards. Bouncing the microwaves doesn't increase the thrust, but reduces it. The waveguide isn't 100% efficient, so photons are leaked from both ends. If the entire wave was expelled from 1 end with no reflections, it would be more efficient.

Like a modern dot com applying thousands of megabytes of a multitude of programming languages to print hello world, it's a case of success achieved by extreme overhead & enough complexity to sound convincing. So no. You can't buy a self driving car, there are fewer jobs than there were in 2007, newtonian physics hasn't been broken, but who knows if living in reality is still necessary.
Posted by Jack Crossfire | Aug 01, 2014 @ 08:57 PM | 2,362 Views
Going to Bakersfield would have been a total disaster. Don't think there was ever another time when something more disastrous was about to happen, outside of politicians. There were definitely crooked times & times which felt like they caused disaster, but nothing that extraordinarily, overtly, outright bad. Fortunately, wasn't wrong about marriage.
Posted by Jack Crossfire | Jul 29, 2014 @ 10:59 PM | 2,430 Views
After Orbotics & Anki comes Sifteo. They weren't bought by Google, but it was another play to make a quad copter into an intelligent toy. Intelligent toys are still probably the most viable market for quad copters.

There are many videos naming the sensors in the Sifteo cubes: acceleration, RFID, & touch screen, but few complete examples of a game being played. There are some home videos with some employees explaining the games, with very little screen footage. As with Goog glass, Goog Tango, & chromecast, the intent is to provide a platform with an assortment of sensors, some industrial design, some really bad marketing, & rely on 3rd parties to program it.

Some game examples:

There were 3 seconds of 1 cube being used as a claw to grab objects in 3 other cubes by tapping the claw cube on the other cubes. Some games use each cube as a scrabble game with bigger pieces. There are centipede games where the player has to move the field rather than the centipede. There are some guys who became fascinated with developing for them.

The buyout is of course not about the cubes, but an easy way of getting people with experience designing toys to design a flying toy platform. The Sifteo cubes require a base station, so the indoor quad copter game will probably use a base station for navigation.
Posted by Jack Crossfire | Jul 28, 2014 @ 07:47 PM | 2,372 Views
There was a long time rumor that the Donner party, or at least the leftovers, were buried in Dublin. The rumor was false. Elisha Harlan is buried there, but was only a grandson of George Harlan, who led the 1st wagon train on a harrowing journey through the Hastings cutoff. Various pioneers who made the 1st crossing of Hasting's cutoff in the same season as the Donner party are buried in the valley, but none who actually got stranded in Truckee.

Little was made of Brigham Young using the Hastings cutoff to reach Salt Lake. It was just a footnote in all the literature. Truth be told, it was a big deal. The route was initially proposed by Lansford Hastings, a confirmed con artist. He hoped to create an independent nation in Calif* with him as leader by getting Americans to migrate through this faster route.

In reality, the path through the Wasatch mountains was virtually impassible. The Donner party traversed it just a year before BY & it contributed to their deaths. Kiwipedia tells a harrowing tale of boulders having to be moved, harrowing canyons, brush having to be cleared, wagon wheels having to be locked because of the steep inclines, & only 1.5 miles per day of progress.

BY followed this same ill fated route, was the 1st one to lay eyes on it after the Donner party crossed & perished, but was apparently much better equipped with 143 men instead of 30. His party followed the tracks of the Donner party & benefited greatly...Continue Reading
Posted by Jack Crossfire | Jul 27, 2014 @ 02:49 PM | 2,355 Views
Surprised to find was shut down. Just like the fever that accompanied the initial start of Gimp & GTK, another piece of the golden age has passed. It was 1 of many bought by VA Linux during that golden age. It & all the other VA Linux assets were eventually bought by dice, "specialized websites for select professional communities". If you got a project on it before it was shut down, you're now immortal.

They claimed a loss of traffic. There's definitely a loss of interest in source code level, freely downloadable PC software. A new generation that would have followed the GPL path has gone to paid mobile apps instead, while the generation that started has moved on to middle management & families.

The itunes/play system that Apple started was such a revolutionary way of automating software releases & getting paid, there was no way everyone was going to keep going GPL, even on the merits of free source code "benefiting the community". No-one in 1999 thought the evil commercial software industry would find a way to lure developers back out of open source, but convenience & ease of getting paid ended up being the factors that did it.
Posted by Jack Crossfire | Jul 26, 2014 @ 06:45 PM | 2,370 Views
Rope Swing Zipline - NFL Stadium (2 min 38 sec)

Note how the brushless gimbal equipped quad is no longer the center of attention but a transparent tool as normal as a shoe. In just a year, perfectly stable shots from a floating vantage point a few feet from people dangling in mid air that were never seen before are the normal look of video. Part of it was still an illusion because the stadium was empty of anyone it could crash on.

Unfortunately, you can't stay a cool zip lining 20 year old forever, even though no-one ages on the goo tube.
Posted by Jack Crossfire | Jul 25, 2014 @ 11:48 PM | 2,641 Views
Posted by Jack Crossfire | Jul 24, 2014 @ 07:42 PM | 2,994 Views
The follow cam proof of concept is everywhere & nowhere. The problem is flying was only possible when someone else paid for it. Without any money, it's going to be a short effort.

Being a spare part application, 1 of the tiny servos from the Blade CX2 will be glued to the NTSC camera for pitch control. The Blade CX2 was an intriguing toy which never flew because it required a lot of trimming & its brushed motors were terrible.

The servo cam will be attached to the underside of a microquad. The microquad needs a better means of vibration damping, to use beat up propellers. It's getting near adding ballast like everyone else.

The camera will track a dual marker baton. The entire baton may actually be colored, making for a dumbbell marker that is easily isolated.

2 voltage regulators for powering the 5.8Ghz transmitter, & servo will be attached to the microquad. The camera is powered by the 5V flight computer rail, which originally was to power the PX4flow. The camera broadcasts video to the ground station, which combines it with telemetry to get a position for broadcasting back to the flight computer over bluetooth.

The ground station needs to record the video as well as scan it for position. The only spare part is a laptop. Since the follow cam's mane mission would be capturing running, the need to carry a laptop has been the showstopper.

For the 5 guys with jobs, the long reigning Odroid is finally dead. You need to be buying the Minnow Max. Unlike the boat, the Minnow has a dual core 1.3Ghz. It should be far ahead of the quad core ARMs. Personally have gotten nothing but grief from quad core ARMs.
Posted by Jack Crossfire | Jul 24, 2014 @ 12:12 AM | 2,333 Views
Quad copter hoverbikes, autonomous follow cams, drone pizza delivery, self driving cars, self balancing motorcycles, & $5 trips to space are just around the corner, now. They've been just around the corner for a while, but now they're really just around the corner. Another impossible claim seems to come out every day, making all the stuff from last week obsolete.

The promises are getting bigger & bigger, but they're all extrapolating from the same old technology which hasn't improved in years. We've all been in a company that's trying to get bought out or have an IPO. They report every smallest, most tentative sale for the next 10 years, every possible source of revenue they could possibly make for the next 100 years, all as their current quarter's income.

That's what the current promises feel like, but every new promise extrapolates more than the last one. Merely breathing now nets $1 million in kickstarter donations & a belief that unicorns are going to ship in 3 months, just because unicorns breathe.

The current movement is feeding on the positive psychology movement that began in 2005. What began as the positive psychology movement mutated into refinancing ever less amounts of property into ever more amounts of money & that now has mutated into a requirement that every invention be believed, no matter how preposterous.

No-one asks questions anymore. No-one is allowed to have doubts. Pictures have nothing to do with what's...Continue Reading
Posted by Jack Crossfire | Jul 22, 2014 @ 08:10 PM | 2,839 Views
Collinn Quinn demos (54 min 23 sec)

With so many unemployed programmers, it was only a matter of time before someone would try to free the bits & upload some videos. We'll see how long this lasts until the indy movie studios crush the rebellion. These videos covered what currently are the 3 mane copters, in a rare appearance without any editing.

Collinn was a lot more down to Earth than when we was at DJI. It was still definitely a choreographed team demo, looking a lot like a Vegas trade show.

As with personal experience, the 1st attempt at a programmed flight was a hair raising flyaway & it took 2 attempts for it to work. He assumed it was a conversion from centimeters to meters, but the team never said what it was. When it did work, the droidplanner eye candy obscured what was happening, but you could barely see that it wasn't able to fly in a circle, in the wind.

They may have removed the circle from the flight plan because it wouldn't work. The user interface never showed it.

The eye candy made Droidplanner worthless outdoors, but it did show the potential for easy mission planning. It just needs a way to play back a simulation before flying, in order to avoid take 2.

DJI Phantom Vision demo (46 min 41 sec)
...Continue Reading
Posted by Jack Crossfire | Jul 22, 2014 @ 04:56 AM | 2,521 Views
20 years after it began, someone finally figured out how to route the audio output in ALSA back to an audio input. The sequence for recording from a Firefox player is:

modprobe snd_aloop
FLASH_ALSA_DEVICE=hw:1 firefox
Make Cinelerra record Loopback #1

The problem is the recording samplerate has to be the same as the playback samplerate. It has to be determined by trial & error or by playing the source in another player besides Firefox. If the player doesn't have a way to send audio to another device besides hw:0, you're screwed.

There is theoretically a way to do it with a line like pcm.!hw in the /etc/asound.conf file, but this never worked. pcm.!hw is supposed to replace any device called hw with what's described in the file.

Some time in 2012, someone wrote the snd_aloop module. The last ALSA release which could be easily compiled without upgrading the entire kernel was 1.0.25 & it has the snd_aloop module.

It creates a virtual soundcard for playback which really routes the audio to the recording handle of another virtual soundcard. aplay -l lists the virtual soundcard as

card 1: Loopback [Loopback], device 0: Loopback PCM [Loopback PCM]
card 1: Loopback [Loopback], device 1: Loopback PCM [Loopback PCM]

which translates to hw:1 for Firefox. They never quite figured out a naming convention, back in their dorm rooms in 1997, so there's a hodge podge of hw:1, Loopback #0, hw:1,0 naming conventions. Playing audio to Loopback #0 makes...Continue Reading
Posted by Jack Crossfire | Jul 20, 2014 @ 08:52 PM | 2,318 Views

It was finally time to use some of the electronic bits from years ago.

There's a 5.8Ghz receiver labeled rx5808. It needs 3.5-5.5V. A channel pin is grounded to select the channel.

A USB capture board is wired to power the receiver with 3.3V.

...Continue Reading
Posted by Jack Crossfire | Jul 19, 2014 @ 10:13 PM | 2,292 Views
Searching far & wide for a marker system which might enable a follow cam revealed this:

Previous experience showed the classic international pink was the best chroma keying material & a doughnut was the easiest shape for current cameras to isolate. When chroma keying is combined with a unique shape, it gives 2 layers of redundancy.

A 4k camera detecting a full body, international pink, digital camouflage suit would be ideal.

Next would be digital camouflage markers, but given the reality of only having enough money for 640x480, we have only

...Continue Reading
Posted by Jack Crossfire | Jul 19, 2014 @ 02:23 AM | 2,348 Views

Despite many flips & spinouts, it still managed to do it in 1h59m. Fastest mile was 8 minutes. Stopped the clock only for the battery change, but counted all the red lights & restarts. The battery died at 12 miles. Its speed rapidly dropped to 9m33s/mile. The new battery was back to 8m50s/mile. Battery life was 1h50m, giving a current consumption of 490mA with the headlights on. The right headlight constantly got bent down.

The human was powered by a salad & some sugar sticks. It made quite a bowel movement, but no crash.
Posted by Jack Crossfire | Jul 17, 2014 @ 07:43 PM | 2,280 Views

So the bottleneck in making a viable follow cam is a super high resolution camera transferring realtime video to a really small computer where it's scanned for small, finely detailed markers. Interfacing HDMI has become the requirement for getting any kind of realtime video. You can probably get it into the raspberry pi's CSI bus. The CSI bus is a parallel bus of differential pairs. The pi isn't fast enough to do anything with it.

The only way to do the job is an FPGA implementation of the marker tracker, doing the full SDRAM, HDMI injestion, & marker detection in hardware. For all the hype about software, OpenCV, & running Pixhawk on Linux, hardware is still the only way to get anywhere near realizing the promises the UAV industry is making.

It's surprising no-one is focusing on hardware implementations of higher end object detection. There is a slow increase in hobbyist attention to FPGA's, but only for controlling LEDs, software radio, or very limited chroma keying from a low end VGA cam.

For all the different platforms they're trying to get Pixhawk to run on, they might as well port the autopilot to hardware. Everyone wants to be bought by Google & Google is a software company, so there's not a lot of thinking outside the box.
Posted by Jack Crossfire | Jul 16, 2014 @ 04:38 PM | 2,553 Views

Another rover test ended when a motor fell off. Everything not threadlocked is being revealed when it falls off. Nothing was originally threadlocked because it wasn't supposed to work. It did 8m40s per mile when it worked. It was extremely hard to keep up. It flipped over twice on pavement cracks.

Accessing the offending bolt revealed all the sensors for the 1st time. The alignment is so critical that it didn't work when reassembled. Another sensor for the left motor had to be shifted ever so slightly closer. Fortunately, the error was a visible amount.

At 8.8V, with no headlights, the right motor now used 60mA more when going forward than when going backward, without load. Left motor used 20mA more when going forward.

At 15V, with no headlights, the no load current of a single motor was previously 250mA. The left motor was now 230mA going forward & 200mA going backward. The right motor was now 430mA going forward & 220mA going backward.

After reseating a bearing & another sensor reconfiguration, the right motor used 250mA going forward & 260mA going backward. This should have a significant impact on range. Another test run revealed it may now be too fast. So any direction problem is decidedly the mechanics rather than the sensors.