grandpasrobots-blog - Tumblr blog

grandpasrobots-blog · 7 years

Text

Why There Is No Such Thing as Negative Energy

I listen to NPR all night and last night they started going on about that dark energy BS!

There is no such thing as negative energy. It is nothing more than a mathematical construct created by people who have no understanding of how relativity works.

The same is true of dark matter. The rate of expansion of the universe is not increasing. And there is no hidden matter making the galactic arms act like spokes and not like planets.

Matter creates the space it is in. Start taking matter out of the universe and it gets smaller. Take it all out and space disappears. And as the universe expands the matter creating the space becomes more dispersed so that the space between any two points actually gets smaller. That is why the Type 1A supernovae (a standard candle) don't appear to be as far away as they should for the red shift they have. The red shift is a measure of the speed of expansion. So if things are not as far away as they should be then they must be accelerating right? Wrong. the space got smaller.

The same effect causes the spoke like behavior of galactic arms. In rotating masses like galaxies or solar systems or satellites around the earth the closer a satellite is to the center of gravity the higher its angular velocity. That means the galactic arms should spread out into a cloud. But they don't. Why? Because the circumference of the orbit of any given object orbiting the galactic center is not what it appears to be. The closer it is to the galactic center the more the space it is in is dilated by the super massive black hole at the galactic core. Being dilated by the black hole is making more space closer to the center. The circumference of an inner orbit is the same as the circumference of an outer orbit. So the arms don't fall apart.

It is interesting to note that some recent astrophysics articles have theorized about why the size of a galaxy seems to be proportional to the size of the super massive black hole. Maybe that's the only stable arrangement.

It is also interesting to note that in spite of the fact that definitive experiments have proven that dark matter does not exist the physics community stands behind it instead of believing a theory (relativity) that has never been shown experimentally wrong by even a tiny percentage of a micro gnats ass.

0 notes

grandpasrobots-blog · 7 years

Text

AI and the FAA: The Secret Life of Drones

When the FAA put out the new drone rules everyone was up in arms. They perceived them to be a long list of things they could never do. Here is the full text with all the arguments, justifications, and arbitrations that resulted in the new drone rules.

https://www.faa.gov/uas/media/RIN_2120-AJ60_Clean_Signed.pdf

I read most of it and skimmed the rest. You know how boiler plate is. But here are a couple of high points that have to do with Artificial Intelligence. First, don't run into anything else in the air (see and avoid - page 18). Second, you can do anything you want if you can prove the tech (waivers - page 36). Then there are concerns about the national airspace, privacy, and safety on the ground below air operations. A national database of air spaces that individuals can exclude their property from would be a start. Or maybe it could be done the other way by registering places where drones CAN fly and take pictures or videos. But either way by using required onboard systems that can override higher level commands safety and privacy can be assured. It would require a GPS and an accelerometer in the camera to sense tilt, zoom, and pan. Some might say this is some kind of outrage but the fact is that airspace must be respected and if we want to fly in it we have to comply. There really is no alternative if you expect to operate as a legitimate business.

But that isn't what this is about. This about autonomous vehicles and how in the very last rule the FAA will give you a waiver to try anything you want. The 'See and avoid' rule for in vehicle pilots is at the heart of the 'Line of sight operation only' rule. The problem is twofold. First, pilots do not generally have access to immersive technologies that are just like being there. Second, communications can be lost and the system is effectively out of control. The first problem can eventually be solved with remote viewing technologies but the real solution is autonomous operation. And autonimity is the only real solution when there is a loss of communications. The problem is that making the vehicle autonomous is a little easier than making a car autonomous (but with much higher costs of failure) but it could not possibly carry all the whirlygigs it takes to do the job. And they would also create a lot of aerodynamic problems. They certainly don’t help the look of the cars much.

The answer is the same as what it must ultimately be in these ugly cars. The AI must learn to use simple vision and its knowledge of the world to create a "World View". And it must constantly update this world view and manage "Goal Directed Behavior". It must also deal with loss of situational awareness. These are concepts I learned from the work of James S. Albus in his book "Brains, Behavior, and Robotics" and the movie “Top Gun” and am using in the grippcad based Carbot. More about that later with free downloads and a link to kits.

So let's figure this out, get the waivers, and have some fun with this while possibly making a living.

Here are a couple of related links. I think my computational theory of mind in grippcad is better than the one at this link but we will see. Grandpa could just be nuts.

https://en.wikipedia.org/wiki/James_S._Albus http://www.latimes.com/business/la-fi-drone-rules-20160829-snap-htmlstory.html

0 notes

grandpasrobots-blog · 7 years

Text

A New Beginning

When the first integrated circuits came out we were all fascinated with what had just happened to electronics. Suddenly we could build complex hierarchies of transistors that could be arranged not only as logic but also for storage of states. And we could build dedicated circuits for a variety of logical operations. Soon we figured out the most useful operations and implemented the corresponding circuits as gates. Shortly after that we figured out how to group gates into registers, counters, comparators, shift registers, and arithmetic logic units. Soon we figured how to use machine opcodes to select the hardware block to operate on the registers and suddenly we could implement the von Neuman architecture in solid state electronics. Supercomputers became desktop machines. The world would never be the same.

Overnight computer stores popped up. But they weren't like the ones we saw in the 90s. These computer stores only sold parts. They had no motherboards or fancy glowing cases or big switching power supplies. It was just parts. It was the buyers responsibility to wire up the memory to the microprocessor and to use dip switches to program it. There wasn't even a computer you could buy to write the programs for the new microprocessor. So people bootstrapped a solution that eventually evolved into DOS. Bootstrapping is an interesting process. Here is roughly how it went. First someone used switches to write a machine language program that did basic I/O. Then they added the functionality to assemble a program from simple text statements into machine code. Using the newly created assembler they rewrote the assembler in assembly. Then they used the new assembler to rewrite everything thus far in assembly. Now we had a bootloader and a language to create a program the bootloader could load. This program was called an operating system and the loader was called BIOS.

The bootstrapping didn't end there. Someone decided to write a program in assembler that could compile or execute a program written in a higher level language. If it executed it directly it was called an interpreter. This was the BASIC language. But of course this compiler or interpreter written in assembly was not very capable. So they rewrote the compiler in the higher level language it was designed to compile. Then they rewrote the operating system in the higher level language also and we had CPM, DOS, Windows, etc.

But something happened along the way that required programmers to look at the platform differently. Operating systems had to be protected from misbehaving programs and unauthorized users. Back then systems were centralized and developers wanted to be able to make desktop computers into multuser servers. This was part of the split between IBM and Microsoft. IBM wanted OS2 (Windows NT) to be multithreaded but not multiuser. Microsoft wanted it to be multithreaded and multiuser. So Microsoft took their copy of the source code and went their own way.

Microsoft did not foresee that Windows NT would become a client/server machine. The old multiuser model was out the window and Microsoft had to find a a new way to charge for the OS. The answer was to make people choose between attached seat and simultaneous network client connection licensing. Great. More crap in the inventors way.

Of course we beat that greedy and intrusive licensing scheme with n-tiered client/server architecture and UDP. And legally too.

We went from centralized processing with terminals to distributed computing on desktops. Because most terminals on centralized computers were connected by low bandwidth serial lines they could not compete with the graphics on a desktop computer. And premodern networking was born.

Then the internet happened. Web servers and browsers became ubiquitous. Bandwidth went up. That document language we made fun of that the secretaries used became the language of the web. The cloud formed. And within a few years we were back to centralized computing and terminals. Except this time the terminals were smart terminals called browsers.

But sadly somewhere along the way the electronics inventor lost the connection with the hardware. The platforms needed to be protected from vulnerabilities caused by their own need to expose themselves. Inventors no longer owned the devices or the software they bought. Operations have become dependent on connected services and fail miserably without them. Operating systems became vast and complex. Development environments got expensive and incomprehensible to the average inventor. And it became impossible to do real time control on modern operating systems.

But most irksome to the maker community is the concept of software as a service. We like to own what we buy. We like to chop it up and make something else or make it better than it was before. But the government is shoving SAS down our throats. I found this out recently in a sales tax dispute with the state of Texas. They insist that the source code to the software I was paid to develop for oil well analysis belongs to me and not my customer. Of course that could be because they want to charge sales tax.

Now we have a new playing field with Arduino, RaspberryPi, FPGAs, and SOCs. And with the new sensors and other peripherals at the prices they are the playing field looks very different. The development environments of the future will bear as much resemblance to text based coding as command line prompts do to an iPhone interface. Automation will be managed with proven logic techniques long used in electronics. Real time systems will work.

But the best part is that we get to program to the metal again. If we can think it we can do it.

So stay tuned for what promises to be a wild ride into automation, robotics, and artificial intelligence.

0 notes