Pages
▼
Wednesday, October 27, 2004
October surprise
On this day, the Red Sox won the world series after 86 years of unsuccessful attempts, a total Moon eclipse could be seen that night, the Huygens-Cassini probe made the closest ever picture of Titan and Iris was born.
Tuesday, October 26, 2004
Thursday, October 21, 2004
Wednesday, October 20, 2004
Questions you have to ask yourself
At LL2, Todd Proebsting had this very insightful talk on innovation in programming languages. His first slide is about a extract of an orbituary of Hamming
The whole video of his entertaining but insightful talk is at : http://web.mit.edu/webcast/ailab/mit-ll2-s2-09nov02-80k.ram What is amazing is that his talk can be applied to so many different problems areas not related to programming languages that it really strikes a chord with me. Some of the gems:
On disruptive constraint solvers:
Richard Hamming’s three questions for new hires at Bell Labs:
1- What are you working on?
2- What’s the most important open problem in your area?
3- Why aren’t they the same? (Ouch!)
“You and Your Research” --- Richard Hamming (1986)
The whole video of his entertaining but insightful talk is at : http://web.mit.edu/webcast/ailab/mit-ll2-s2-09nov02-80k.ram What is amazing is that his talk can be applied to so many different problems areas not related to programming languages that it really strikes a chord with me. Some of the gems:
It's Bell Labs, we hired you because of your judgement but you are obviously not using your judgement because you are not working on the most important problems in your field
I am glad to hear you're on it
Undo is hard
(map fn L) vs. while (*d++ = *s++);
It's almost never a regular expression
Parsing is too hard
On disruptive constraint solvers:
You give up, which is what I did
Prolog is fine but it does not solve other problems
Objects are fine, I am giving you a 1000 of them
Monday, October 18, 2004
Solar System Weather Report
Thursday, October 14, 2004
Bandelets to bar code
The company LET IT WAVE is developing a technology close to Curvelets and Contourlets called Bandelets. While the JPEG 2000 standard is an improvement over the current JPEG format, the bandelets seem to be an improvement over JPEG 2000. Here is the same photo truncated to 500 bytes so the information could fit into a bar code
Pretty impressive.
Pretty impressive.
Wednesday, October 13, 2004
Entering the DARPA Grand Challenge Race
After much time looking into it, we are ready to make this hobby a more serious hobby. I created a blog for our entry into the DARPA Grand Challenge race. Not much detail can be found there because with a small team, we'd rather concentrate on the most important thing, the vehicle and its autonomous driving system. Why Pegasus Bridge ? Because, on all accounts, we don't think we are building a car....
Tuesday, October 12, 2004
Peak Digit
It looks like as soon as the thriteenth bar will be implemented, we will be already full and looking to have a fourteenth. Yes, it is not justoil anymore, we now have Peak digit.
Monday, October 11, 2004
Have you ever seen a spacecraft being tested ?
Here is one: The ESA Venus Express. The reasons these guys have white gowns and hairnets is because you don't want to contaminate the spacecraft with particulate matters. Since the Venus Express is going to Venus, it is likely to feel very warm since it will be closer to the Sun during its mission. It is therefore imperative that any of the Multi-Layer Insulations cloth and other outside thermal protection materials properties do not degrade because of contamination. Any property changes of the outer layers will directly have an impact on the temperature inside the spacecraft.
A Master Builder Passed Away
Max Faget passed away this week-end. Here is a quote I like from what is being said about him:
His designs included the escape pod of the Saturn V rocket (the metal tube at the front of this picture is a rocket with three small nozzles) and the wake shield facility,
Ad astra Max.
Seeing things from a different angle was another strength, and he sometimes made dramatic demonstrations of this.
His designs included the escape pod of the Saturn V rocket (the metal tube at the front of this picture is a rocket with three small nozzles) and the wake shield facility,
Ad astra Max.
Sunday, October 10, 2004
Friday, October 08, 2004
GoogleLabs, Graham and NewLisp
John Wiseman has three interesting links:Peter Norvig at Web 2.0 on search clustering, Paul Graham on Arc and an older entry on NewLisp. Good stuff.
Seven oil platforms were destroyed by Ivan
This is the type of news we've not heard about when hurricane Ivan was hitting the Gulf coast. But it looks like seven platforms were detroyed. I just added this information to wikipedia. At $52 a barril, we are approaching a perfect storm. This is noticeable because the actual effect of the destruction of these platforms on the daily consumption in the U.S. is actually very minimal.
Thursday, October 07, 2004
A-ha moment, part I
Through the course of learning, one always tries to make sense of the learning material through story-telling or through an obvious internal representation. The a-ha moment is when you realize that you now have found this internal representation... and you can't get it out of your mind. It may be triggered through an awesome graph or a simple key sentence that makes the connection between two dissimilar subjects. Recently, when I was reading the literature on on-line parameter estimation using the Unscented Kalman Filter, I was having problems making a comparison between the unscented transform and previous methods such as the Extended Kalman Filter.
Featured sideways is the graph that explained it all. Similarly, one cannot understand the slaughter of Napoleon's army during the Russian campaign until ones contemplate this map that connects the thickness of the line with the number of surviving troops. Similar very good examples can be found in this gallery of Data Visualization.
Tuesday, October 05, 2004
Intelligent Thermal Vacuum Chambers
In this invitation to bid from ESA, they seem to have a need for a way to accelerate thermal testing. In their words,
Maybe a way to do this is through better analysis while the testing is being performed ? Maybe an idea would be to allow for the chamber to be intelligent about the actual thermal cycle the payload has to go through ? Imagine that, Intelligent Thermal Vacuum Chambers ...
Modern spacecraft require lightweight solar arrays capable of withstanding either at least a thousand very deep thermal cycles in Geostationary orbit (GEO), or several tens of thousands medium thermal cycles in low earth orbit, or high range temperature thermal cycles for space missions to the sun vicinities (e.g. Bepi Colombo, SOLO, etc.). Extensive tests on solar array samples have to be performed in order to ensure that mission requirements will be met. The objective of the activity is to procure an accelerated thermal cycling facility with the required characteristics to fully cope with the thermal cycling test conditions defined for LEO, GEO and close to the sun missions, with the additional ability of performing short duration cycles, so the testing time for qualification and evaluation programmes can be considerably reduced.
Maybe a way to do this is through better analysis while the testing is being performed ? Maybe an idea would be to allow for the chamber to be intelligent about the actual thermal cycle the payload has to go through ? Imagine that, Intelligent Thermal Vacuum Chambers ...
How can a 32,000 lb truck be stopped by a bush ?
Just make it driverless. A more accurate analysis can be found in this document relating Terramax's entry into DARPA grand challenge of last year.
Nice little exercise to go from sensor reading to a GUI
The nice thing about python is the quick prototyping capability to do this sort of thing.
NOAA-N-Prime Satellite Mishap Investigation Report Released
The NOAA-N-Prime satellite mishap investigation report has been released. This is a follow-up of this entry. Please note the following in the report:
In my view, the conclusion of the report points to one thing, no matter how many "Q/A" people you have on-board, if the organizational structure is not working, then you cannot expect a Quality Band aid program to help in anyway.
Errors were made by other team members, who were narrowly focused on their individual tasks and did not notice or consider the state of hardware or the operation outside of those tasks(Page 9.)
In my view, the conclusion of the report points to one thing, no matter how many "Q/A" people you have on-board, if the organizational structure is not working, then you cannot expect a Quality Band aid program to help in anyway.
Monday, October 04, 2004
The September issue of "La Lettre de la SCM" just came out.
La Lettre de la SCM just came out
In "Eoliennes", there is a mention of wind turbine failures given as a very low probability by experts when, in fact in the course of four months, two have been destroyed (because of the wind!) What is fascinating is that with the increase in power ratings of wind turbines, we are likely to see these probabilities even go up. Indeed, three hundred of these turbines could replace one nuclear power plant. This is not to say that electricity would be cheaper (currently the wind power industry is subsidized by the government) nor that it would help France in fulfilling Kyoto's treaty CO2 levels promises since both energy sources do not produce any.
"La théorie n'est pas au point" mirrors my previous comment on orders of magnitude in the prediction business when it comes to the environment.
In "Identification", there is a mention made of the fiasco in using biometrics issues when it comes to identifying people and that smell may actually be a better classifier than face recognition. This is a good idea, as it stands, dog smell is already looked at to detect some cancers. Smell might actually be a better detector of intent than the currently flawed idea that databases are going to protect us. As a traveler, do I care more about a seating next to
1) a serial killer or,
2) a bomb ridden "peaceful" person ?
The database says 1. I say 2. In a related news, Smell study has netted some researchers the Nobel prize today.
The rest of this september issue of the SCM Letter is a very good and entertaining read (if you read french that is.)
In "Eoliennes", there is a mention of wind turbine failures given as a very low probability by experts when, in fact in the course of four months, two have been destroyed (because of the wind!) What is fascinating is that with the increase in power ratings of wind turbines, we are likely to see these probabilities even go up. Indeed, three hundred of these turbines could replace one nuclear power plant. This is not to say that electricity would be cheaper (currently the wind power industry is subsidized by the government) nor that it would help France in fulfilling Kyoto's treaty CO2 levels promises since both energy sources do not produce any.
"La théorie n'est pas au point" mirrors my previous comment on orders of magnitude in the prediction business when it comes to the environment.
In "Identification", there is a mention made of the fiasco in using biometrics issues when it comes to identifying people and that smell may actually be a better classifier than face recognition. This is a good idea, as it stands, dog smell is already looked at to detect some cancers. Smell might actually be a better detector of intent than the currently flawed idea that databases are going to protect us. As a traveler, do I care more about a seating next to
1) a serial killer or,
2) a bomb ridden "peaceful" person ?
The database says 1. I say 2. In a related news, Smell study has netted some researchers the Nobel prize today.
The rest of this september issue of the SCM Letter is a very good and entertaining read (if you read french that is.)