Currently, I am totally in love with the PyBoard, available form AdaFruit in the US. The PyBoard is the reference implementaiton of MicroPython, which is Python 3.x ported to microcontrollers. Very nice implementation of Python -- a (very) few differences from CPython because you are running on a limited amount of physical memory on bare metal, but a very, very nice version of Python. I am telling all my robot-building friends about it every chance I get.

That said, the Arduino has a huge community by virtue of having been around a long, long time. You can get a RedBoard (my favorite Arduino clone) from SparkFun for lunch money, so if you toast one, who cares? I recently did a real, live, paying contract using an Arduino Micro, which uses a Mega 324, which a much more capable processor than the Mega328 that is on the basic Arduino boards. SparkFun has a "SparkFun Inventor's Kit" around the RedBoard that gives you a selection of sensors and actuators to plug in, and a project guide. You might like that as a way to get started.

Also... about blowing things up.... don't fear the magic smoke. As a youth I was deathly afraid of toasting my components. That probably had to do with my limited allowance :/ I would have accomplished more had I simply forged ahead, seeing the occasional dead component for what it is: the cost of tuition. The great thing about being able to buy a whole computer for the cost of a burrito and beer is that fear of magic smoke is no longer a thing.

Yes, I like that idea.

Also... perhaps there should be some new form of meta-moderation that is triggered by multiple conflicting moderations. It almost never fails that if I post actual, real-world, data that is in conflict with someone's world-view, I will get both up-modded and down-modded. Sorry, there should not be a "-1, uncomfortable truth" moderation. Perhaps posts that accumulate a large number of conflicting moderations should cause a form of meta-moderation that concentrates on uncovering biases in the moderators.

Getting the software to work is 90% of the effort. Getting the hardware to work is the *other* 90% of the effort. Fixing broken hardware is the *third* 90% of the effort.

Robotics requires great software engineering, great electronic engineering, and great mechanical engineering. Few people are good at all three. No one has time to do all three by themselves for anything but very basic robots. And then there is budget.

The OP wants to do serious robots on the cheap. ROS and Gazebo are the way to do that. Will the simulator model every way the robot can fail? Nope. But, it turns out, that the simulator faithfully models *enough* ways for the robot to fail to keep a serious team of people busy for many years.

As to arms versus roving: IMHO building a satisfying rover is easier that building a satisfying arm. The mechanical design challenges are much simpler. The software can be endlessly complex for either, depending on your goals.

You ask if pros have the experience to avoid most newbie hardware mistakes. Yes. They got that experience by making those mistakes themselves, or watching their friends make those mistakes. (My standard advice to newbies is that your first robot should have a low enough mass that a simple software bug will not punch a hole in the drywall of your living room.)

As to where does the trouble-shooting effort go, after 10 years of serious hobby robotics I would say that someone that says they spend 90% of their time debugging hardware problems is interested in hardware and is pushing the development of hardware and probably runs simple software, often with the robot on a test stand. Someone that says they spend 90% of their time debugging software is taking on a bleeding edge software problem and is running on off-the-shelf hardware. It's all good. It's how we make progress.

Get a beefy workstation, and bring up ROS using the Gazebo simulator. You can define your own robots, and you have a full physics simulator. You might look at the Fetch Robotics web site for a demo that you can download. The demo gives you a full model of a Fetch, and a demo script that you can start hacking on. This is exactly how the developers at Fetch work up their code before deploying on actual hardware.

So not only is it possible to do serious, state-of-the-art, robot software development in a purely simulated environment, if you go to any serious robot research lab in academia or industry, you will have to show your code working on the simulator *before* you are *allowed* to run your software on real hardware.

ROS has pretty much taken over as the framework for academic research except for places like CMU and MIT that have many years invested in their own robotics software frameworks. Even at those places, there are labs switching over to ROS depending on the underlying hardware. In industry, you find companies prototyping in ROS and replacing open source modules with "secret sauce" modules to market differentiation as they ramp. ROS is also where serious hobbyists are going now. So, if you want to get into a robotics graduate school program, or into robotics industry, learn ROS, do something original, and put together a good demo screencast.

So..... you will not run the Gazebo simulator on a RasPi. Build the beefiest X86 Ubuntu workstation you can afford, and be *very* careful about choosing a compatible graphics card. Go to ROS.org to find specs. X86/Ubuntu is the reference platform and development platform for ROS. There are ports to lots of other platforms, and people *do* run parts of ROS (not Gazebo) on RasPi's. Porting ROS is *not* a beginner's project. Take the easy route. Every. Time. (For now, things continue to get better.....)

Came here to say that. I absolutely recommend the Murderous Maths books. Other good books: The Number Devil, and The Life of Fred. Those were all top favorites of my daughter at that age, who considers herself a fan of all things mathematical.

It is also good for math-loving kids to spend time with other math-loving kids. Here on this side of the pond you can find Math Circles in some communities, which I take it is an idea that the Russian immigrants brought over with them a few years ago. Also, my daughter benefited tremendously from attending the MathPath camp, maybe you can find something similar over there. (Although MathPath has a few international attendees every year. See mathpath.org The first year my daughter went, Katherine Wolfram (yes, her dad is Steven Wolfram) was there, and I can't think of a better recommendation than that.)

That's called Williams Tube storage. Several early computers used it. One of my Computer Engineering profs started on those, then worked on a computer that used mercury delay line memories, a big improvement in storage density. You took a long pipe, formed it into a serpentine that fit the memory cabinet, put an ultrasonic transmitter on one end, and a receiver on the other, forming an acoustic delay line. You regenerated bits, or flipped them, as they came around. Of course, you had to wait for the bit you wanted to fly by.... Later he got involved in some new-fangled stuff called "core", and many of his graduate students went on to be big deals in magnetic core memory design. The last of his graduate students worked on magnetic bubble memory for their thesis work.

Well, autonomous fighting robots have been tried. It isn't very interesting to watch. At. All. Three minutes of watching two robots try to find each other is just not gripping. I work with autonomous robots a lot. Watching autonomous robots attack a 1-on-zero challenge course is something that is usually only interesting to other builders of autonomous robots, because the action is so freaking slow and the robots look feeble-minded. Only another builder can appreciate how difficult it is to make a robot do what appears to the non-initiated to be a very simple task.

There are autonomous sumo-bots. Google for some video. Other than pushing an opponent out of the ring, they aren't weaponized. Those typically move quite fast, so there is action. Occasionally, you get some un-intended action -- I saw a sumo-bot erupt into a pretty spectacular lithium battery fire once. But I don't think you could make a TV show out of sumo-bots.

I'm currently working on a Robo-Magellan bot. I can't imagine wanting to subject that much work to intentional damage. It's freaking hard enough to keep it all working in the first place without someone applying kinetic rage to your electronics.

That is extremely logical. Which in and of itself is enough to keep the US left from ever doing that. But the real reason the left is not trying to amend away the 2A is because they can count votes. And because of what it takes to get an amendment past the house and subsequently ratified, it would never pass. The votes aren't there for either step.

Also consider this: if the left ever proposed an amendment to replace the 2A, it is an admission that the 2A means what it says, instead of what they *wish* it said, which is their current operating procedure. That kind of huge admission totally sinks their cause.

First, recall that the AP test is designed to be worth about 1 semester's worth of credit in a freshman intro course in . So AP CompSci should be thought of as that first intro to programming course. It isn't about deep concepts, it is about testing your understanding of the basics. Memory layout, pointers, memory management, etc, are not first semester intro course topics, and never were. AP CompSci is testing whether you should be forced to sit through intro to programming with people that have never coded anything, or whether you can proceed directly to the courses in topics like memory layout, pointers, and memory management. Don't expect it to be "AP Compiler Design Laboratory for Garbage-Collected, Declarative, Parallel-Processing Languages". That is not the intent.

My daughter took AP Comp Sci last year. The way the current test is structured is sort of interesting. Students are expected to have spent a significant amount of time studying and working with a particular body of Java code before coming into the test. The test itself then asks them to do things like add methods to particular classes to implement particular changes and enhancements. To that extent, it actually has a more practical structure than one might expect. Before doing AP CS, my daughter was quite fluent in Python, and also had done a modest amount of ANSI C and "Arduino C++" for a science fair project. She didn't much groove on the Java, and it still isn't her favorite language, but seeing another language and its approach to things isn't a bad deal either. Learning to read a big chunk of somebody else's code and grok it was also good, and turned out to be a good skill later when she had a summer internship and had to read a big honkin' glop of C++ mush in order to interface her Python code to it. Given her fluency with Python, the AP CS was actually a cake walk -- it really isn't a difficult test. But then again, the final exam for the very first Intro to Programming course for people that have never coded before isn't exactly a killer either for anyone with any previous fluency in any programming language.

From what I can see, the current AP CS exam tests what it claims to test, and gives a decent evaluation of the student's understanding. AP Calc, OTOH, seems utterly broken. I had a chance to talk to both a professor who was, for three years, on the grading committee for the hand-graded part, and also a math prof form Harvey Mudd. The HMC prof said they stopped give AP Calc credit, because a 5 was no guarantee that students knew enough to skip the first semester. The grader commented that he could easily see how -- there is a *huge* range on the AP Calc that can earn a 5. If you smack it, you get a 5, of course, but you also can miss a lot and still get a 5. So from the standpoint of a college trying to understand how much calc you know, a 5 has little meaning. So it seems to me that the AP calc exam is not serving anyone well.

Right on. I worked at Control Data, as a CPU logic designer. The PP's were peripheral processors. The article is full of so much egregiously incorrect tripe I won't even bother to type up a correction. My advice to everyone is to completely ignore the article unless you want your head stuffed full of misinformation.

The best boss I ever worked for was an ex Israeli commando officer. When I say that to most people, they are often a little frightened. But think about it: 1) A commando officer never sends his people in without making sure everyone has a very, very clear understanding of the objective. 2) A commando officer asks what they need and gets it for them or adjusts the plan. 3) A commando officer asks if they can meet the time frame or adjusts the plan. 4) A commando officer truly cares that all his people are as healthy at the end of the operation as at the start.

Anything less results in having to write very unpleasant letters to parents. The same simple steps, especially #1, will go a long way in a software organization, too.

That, and the fees charged are fixed by law. So doing the DROS paperwork without making the sale is a net money loser for the gun dealer. I like to support the local guys as much as I can, and most of them are happy to order anything special that you might want from their distributors anyway.

But diesels can meet emission standards honestly if auto makers include a urea tank. They just fear consumer reaction to having another consumable fluid (that needs to be refilled every 9 thousand miles, or so) and don't want to do the heavy lifting of consumer education. Considering we are a lot closer to a renewable biodiesel fuel than other fuels, it seems like it makes a lot of sense. Of course, that begs the secondary question of whether farm land should be used to grow food or to grow fuel for cars, but that is a secondary debate.

Actually, you won't meet emissions standards without adding a urea tank to process the exhaust on the way to the tail pipe. It isn't magic to take out NOx, it's just that marketing doesn't want to explain why people need to buy another consumable, and have the tank refilled every 8 to 9 thousand miles or so.

Whenever the topic of allowing government or public access to review source code comes up (like with, oh, say, voting machines) I always think of these guys:
http://gaming.nv.gov/index.asp...
and I realize that not of this is as important as gambling (and the collection of taxes thereon).

At least if you judge by how seriously we take access to the code. Just try to deploy a slot machine in Reno without letting someone at the Nevada State Gaming Control Board review your code. Won't happen.