Thursday, February 24, 2005

What YOU've been missing

On february 10h, in his weekly column Bob Cringely described how the whole venture capitalist world will be looking forward to plugging much money in VC supported companies

You may recall that I asked a year ago for venture capitalists, who weren't at that time funding much of anything, to start spending money if only to jump-start the economy. They didn't, of course. So why would they now start to fund every deal in sight? That's simply because their alternative is to give back the money and then some, which they will never do.

In 1999-2000 -- at the very peak of the dot-com boom -- venture capital firms were not only taking companies public at a furious pace, they were just as furiously raising new venture funds -- funds that will shortly be coming to the end of their lives. Throughout the fixed lifespan of these funds venture capitalists are typically paid 1-2 percent of the total fund per year as a management fee. If a VC raises $100 million for a fund with a six-year life, they'll take $2 million every year as a management fee, whether the money is actually invested or not. Any money that remains uninvested at the end of the fund must be returned to the investors ALONG WITH THE ASSOCIATED MANAGEMENT FEE.

Right now, there is in the U.S. venture capital community about $25 billion that remains uninvested from funds that will end their lifespans in the next 12-18 months. If the VCs return those funds to investors they'll also have to return $3 billion in already-spent management fees. Alternately, they can invest the money -- even if they invest it in bad deals -- and NOT have to cough-up that $3 billion. So the VCs have to find in the next few months places to throw that $25 billion. They waited this long in hopes that the economy would improve and that technical trends would become clear so they could do their typical lemming-like jump off the same investment cliff as all the other VCs. Well, we're at the edge of the cliff, so get ready for the most furious venture investing cycle in history.

I'm not saying this is bad, but I do find it ironic.

and then today, Technology Review tells us about the 96% week increase in US capital venture spending

Last week US venture-capital investment jumped nearly 96 percent, to $347.85 million from $177.65 million. In that same time, Wireless investments, alone, jumped more than 250 percent.

Correlation much ?

Wednesday, February 16, 2005

Cognitive Problems and Learning Robots

When I'll be talking at SCM next month, my talk will most likely make the connection between the type of studies currently undertaken in Robot learning and how they can be used towards diagnostics in people with cognitive problems. Generally, this knowledge is applied the other way around. An example of that approach can be found in this article on a Bayesian model of imitation in infants and robots. In it, one can read:

In particular, we intend to study the task of robotic gaze following. Gaze following is an important component of language acquisition: to learn words, a first step is to determine what the speaker is looking at, a problem solved by the human infant by about 1 year of age (Brooks & Meltzoff, 2002). We hope to endow robots with a similar capability.

This is fascinating because one of the reason kids with autism have difficulty learning is that their 'gazing' behavior is not optimal, which in effect stops them from learning by imitation. Another aspect of the connection between the two approaches is how imitation can be a quantifier of Autism at an early age. Video made by families of kids with autism show charateristic behaviors at an early age (9-12 months.) More recent papers show the behavior can be seen earlier at 4-6 months.

Photos of the recovered parts of Starnav

Starnav was a payload located on top of the Spacehab module located in the payload bay of the shuttle Columbia

When closed and ready to operate, it looked like this

When opened, it looked like this

Two parts were recovered in West Texas: the back plate

and the lens housing

Thursday, February 10, 2005

Color is all you need

According to this new technique allowing you to see at night like in daylight conditions, it is very interesting to note that for the human brain, color is really the discriminating factor when it comes to situational awareness. An autonomous car, would need to see and make sense of color.

Friday, February 04, 2005

We want to see everything and make sense of it

When writing proposals, it is always a good thing to look at past proposals and see if there is a trend. In particular, it is interesting to understand the slection mechanism by which the decision is made. Since it takes so much of a researcher's time to write a proposal, reading past debriefings is a good return on investment in answering the question: Will my idea go through the different tests of this organization. For instance, let say you want to contribute to a specific technology for the NRO through the Director's Innovative Initiative. Unlike some other agencies, this one provide breifings from previous years. It is a good thing. Unfortunately, it is all in a pdf format. In order to fully utilize this information, you really need to have some type of templates that uses all the remarks that were made in this briefing. If they are in pdf, you have to retype everything, which is a pain. Fortunately, Adobe allows a pdf-to-text converter to work over the web if the initial document is on the web. It is a nice little tool that should allow me to use the DII's debriefing present in my document so I can write a proposal accordingly. I believe there is a market for this type of application. An application that somehow reminds you of the essential points that need to be put in the document that you are producing. Large and small companies have boilerplates which have seen the test of time. But I think there is a market for small companies that just start entering new businesses.

Wednesday, February 02, 2005

Accurate rendering

It is very rare when a journalist takes time to understand issues and put them in a language that is not technical yet still understandable to the public. Phil Chien seems to have done so in his article published in the Eagle on the experiments that took place during Columbia's final journey.

During space shuttle Columbia’s final, 17-day mission, a large amount of scientific data was relayed to the ground. When the shuttle broke up over East Texas on Feb. 1, 2003, it was feared that all of the science aboard had been lost, in particular biological samples....

Texas A&M University had two experiments onboard — a rat experiment sponsored by the College of Veterinary Medicine and “StarNav,” a prototype star tracker built by students at the Spacecraft Technology Center.

The StarNav team’s goal was to design an inexpensive star tracker to determine the shuttle’s attitude, or orientation. A digital camera was mounted in a tube with a single-board computer. The computer converted those images into readable star charts, and its Lost In Space Algorithm (LISA) program calculated the spacecraft’s orientation. The software has to isolate debris and other unwanted defects in the photos so accurate star patterns can be determined.

There was a problem — the StarNav team had not anticipated too much stray light entering the camera. Before the mission, the team made a list of restrictions for when the instrument could be turned on: no viewing the sun, moon, or brightly lit Earth, or when Columbia was passing through a high radiation region over the South Atlantic Ocean.

You can see this in the following photo (we are behind the white box on the center right on top of the Spacehab module.)

Even the camera that took this shot sees a lot of reflections. We were not expecting that much glare, so we had to change the whole shooting schedule to make photo shoots only when the orbiter was on the night side. This led us into scheduling new star pictures sessions that were originally not on the agenda (after the normal clearance process with SpaceHab and NASA) since most of our shooting sessions seemed to occur when the orbiter was on the day side during most of the mission. We did get a lot of night shots but as Phil says,

Unfortunately, six days after launch, StarNav was powered while going through the high-radiation region. Its hardware was built from non-radiation-hardened components and, not surprisingly, it stopped working.

This high radiation area was the SAA. But it was still a surprise to us.

The StarNav team could not send commands to shut off the unit in time because of other higher-priority communications. Later, a maintenance procedure was sent to the crew to instruct them to use a laptop computer to send commands to the payload.

“Eventually we had to stop, as the camera never was able to perform the duties that were expected from it [after the radiation problems],” scientist Igor Carron said.

Even with those problems, StarNav was a success. The main purpose was for the students to develop a technique that could be used on future spacecraft sensors. They were able to design their star tracker and build it from off-the-shelf electronics, test it ahead of time to ensure it would work in outer space, and make it safe enough to fly on the shuttle.

All together, they had 30 25-minute opportunities to use StarNav, and they received 32 digital images. The onboard computer was able to use those images to determine where the instrument was pointed.

The goal for the experiment was a prototype star tracker, and that was extremely successful. The problems the StarNav team encountered were good problems because they showed the limitations of their concept and where they needed improvements. Texas A&M has used the knowledge from StarNav to develop the Khalstar Star Tracker, which is commercially available.

... For most of the scientists the loss of their life’s work seemed insignificant in comparison with the loss of the seven person crew and the space shuttle.

Scientists who were able to recover their experiments from the debris or had data radioed during the mission are doing whatever science they can.

In a strange twist, the web site on which we described our experiment was, unbeknownst to us, used by members of the recovery team in Florida to figure out that two pieces of equipment found 100 miles north of College Station were coming from our camera.

Philip Chien will publish a book about the Columbia astronauts and their mission entitled "Columbia — The Final Voyage of NASA’s First Space Shuttle."

Tuesday, February 01, 2005

Prometheus: The Paradigm that was busted

Prometheus was seen as a paradigm buster, well it looks like it's going to a busted paradigm instead. Well for one, the news is from NASA Watch

Jupiter Icy Moons Orbiter Mission to be Delayed NASA Watch has learned that the Jupiter Icy Moons Orbiter (JIMO) mission will be delayed. Details will be announced Monday, February 7th during the roll out of the FY06 NASA budget submission.

but then we also learn that it will likely not be nuclear:

It sounded exciting in 2003 when NASA announced that the Jupiter Icy Moons Orbiter mission would be the first nuclear-propelled mission under Project Prometheus. But now it looks like a plan to put them off while NASA focuses on Moon/Mars. Kinky is nice, but if conventional will get to Europa, they’ll take it. Europa may be the last hope of finding other life in the solar system....

If there is anything to be learned from the recent Huygens success, it is that some scientists waited 15 years to see if their experiment would be turned on (not all of them did) but on top of this, because it did not have enough power, the probe could not talk directly to earth and had to rely on Cassini to send messages back. Which it did, but since the probe had conventional batteries, only about 3 hours of data were sent back to Earth. Fifteen years of waiting for three hours of data. Prometheus promised that instead of the tiny wattage generally given to sensors (watt range), scientists would be given large amount of power (Megawatt range) enabling mind boggling science return. It looks like it ain't going to happen.